In soft-sediment habitats there are many examples of species that modify their habitat and thus can be expected to have an important influence on macrobenthic community structure. The large, suspension-feeding pinnid bivalve Atrina zelandica adds complexity to soft-sediment habitats by protruding into the water column and altering boundary-flow conditions and by providing predation refuges and substrates for epifaunal settlement. To investigate effects of A. zelandica density on macrobenthic community composition, we conducted a density manipulation experiment in 4 different habitat types in and around Mahurangi Estuary, New Zealand. Our experiment incorporated a comparatively large spatial and temporal scale: each habitat was separated by at least 1.75 km, and was sampled 3 times over 16 mo. Based on previous work, we predicted that macrofaunal community responses would differ between sites and would be stronger at sandy sites than muddy sites, and that variability in site hydrodynamic and sediment characteristics would help explain differences in benthic community responses to the density manipulation. While these predictions were supported, there was considerable temporal variation in the response. We also made predictions of the response of different aggregate macrofaunal groups to the A. zelandica manipulation (i.e. total numbers of individuals and taxa, suspension feeders, deposit feeders, top 2 cm dwellers, and mobile, short and long-lived species). Whether these predictions were supported varied spatially as well as temporally. To be able to generalise results, larger scale experiments, conducted at more than 1 site and at more than 1 time, are generally considered preferable. Although our A. zelandica manipulation experiment has these attributes, the results have demonstrated that the influence of this large suspension feeder on the associated macrofaunal community is not simple (except perhaps in sandy, relatively non-tidal environments), and illustrates our limited success in 'reducing' the complexity of this system using a field experiment. However, we were able to demonstrate that interactions between A. zelandica, site hydrodynamic conditions and sediment characteristics were all important in influencing macrofauna, rather than there being a simple A. zelandica density-macrofauna relationship. Thus, where multi-species interactions, indirect effects, non-linear biotic/abiotic interactions and threshold effects play an important role, specific experiments may not always lead to generalisable results, simply because the system is too complex.