Increasing recognition is being given to the concept of ‘ecosystem engineers’, which alter the physical nature of the environment and thereby influence other species by means that are neither competitive nor trophic. This paper examines a case study of such effects, in which negative correlations between the abundances of the sandprawn Callianassa kraussi and the eelgrass Zostera capensis imply mutual exclusion because Z. capensis stabilises sediment and inhibits burrowing by C. kraussi, whereas sandprawn bioturbation smothers eelgrass. To test this, we undertook manipulative experiments in which Z. capensis was transplanted into sandflats where C. kraussi was either left undisturbed or eliminated by defaunation. This demonstrated that (1) Z. capensis thrived and expanded in sandflats in the absence of bioturbation but deteriorated and disappeared if C. kraussi was present. (2) In the short term, introduction of Z. capensis reduced densities of C. kraussi, and in established beds of eelgrass, C. kraussi was rare. (3) The mudprawn Upogebia africana was largely restricted to treatments that contained Z. capensis, and its densities were greater in the absence of C. kraussi than in its presence. The presence of eelgrass and the exclusion of C. kraussi also diminished sediment penetrability, suspension of particles and surface burial rates compared to the situation in undisturbed Callianassa-dominated sandflats. These results confirm a mutually negative interaction between C. kraussi and Z. capensis and strengthen the belief that the underlying mechanisms are the antagonistic effects of bioturbation by C. kraussi versus sediment stabilisation by Z. capensis. U. africana seems to benefit directly from Z. capensis but possibly also benefits indirectly from the exclusion of C. kraussi by Z. capensis, since its filter-feeding mode of life requires sufficient sediment stability to maintain semi-permanent U-tubes.