Abstract. Ecosystem engineering-the physical modification of habitats by organisms-can create patches with altered species richness relative to adjacent, unmodified patches. The effect of ecosystem engineering on patch-scale species richness is likely to be difficult to predict from the identity of the engineer, the resources altered as a result of engineering, or the identities of the affected species. Here we develop a simple conceptual model that predicts the effects of ecosystem engineers on species richness based on how the habitat modifications caused by engineers affect primary productivity, assuming a humpshaped relationship between productivity and species richness. We review data from 35 studies that contained 60 comparisons of species richness on patches that had been modified by ecosystem engineers vs. unmodified patches. We found no general patterns in whether species richness at the patch scale was increased or decreased by ecosystem engineering. However, 14 of these studies also contained data on primary productivity on and off engineered patches, giving 30 cases to: (1) test whether the effects of ecosystem engineering on richness depend upon the productivity of the ecosystem, and (2) examine the effect of the engineer on productivity. Matching the predictions of the conceptual model, we found a significant negative relationship between productivity and the engineering effect on species richness when ecosystem engineers increased productivity and a weak positive relationship when engineers decreased productivity. We compare the conceptual model developed here to models predicting the effects of grazing and facilitation on species richness. These results, if supported by further studies, can contribute to our general understanding of ecosystem engineering and have important implications regarding the consequences of the loss or introduction of ecosystem engineers on species richness and ecosystem function across landscapes that vary in productivity.