In a seagrass restoration project, we explored the potential for enhancing the restoration process by excluding antagonistic engineering interactions (i.e., biomechanical warfare) between two ecosystem engineers: the bioturbating lugworm Arenicola marina and the sediment-stabilizing seagrass Zostera noltii Hornem. Applying a shell layer underneath half of our seagrass transplants successfully reduced adult lugworm density by over 80% and reduced lugworm-induced microtopography (a proxy for lugworm disturbance) at the wave-sheltered site. At the wave-exposed site adult lugworm densities and microtopography were already lower than at the sheltered site but were further reduced in the shell-treated units. Excluding lugworms and their bioengineering effects corresponded well with a strongly enhanced seagrass growth at the wave-sheltered site, which was absent at the exposed site. Enhanced seagrass growth in the present study was fully assigned to the removal of lugworms' negative engineering effects and not to any (indirect) evolving effects such as an altered biogeochemistry or sediment-stabilizing effects by the shell layer. The context-dependency implies that seagrass establishment at the exposed site is not constrained by negative ecosystem-engineering interactions only, but also by overriding physical stresses causing poor growth conditions. Present findings underline that, in addition to recent emphasis on considering positive (facilitating) interactions in ecological theory and practice, it is equally important to consider negative engineering interactions between ecosystem-engineering species. Removal of such negative interactions between ecosystem-engineering species can give a head start to the target species at the initial establishment phase, when positive engineering feedbacks by the target species on itself are still lacking. Though our study was carried out in a marine environment with variable levels of wave disturbance, similar principles may be expected to apply to other ecosystems that are ...