I examine the effect of ecosystem engineering on species richness at multiple spatial scales using wetlands created by beaver (Castor canadensis) as a model system. By creating patches containing a different set of species than forested riparian zones, ecosystem engineering by beaver increases species richness at the landscape scale by approximately 33%. I suggest that ecosystem engineers will increase species richness at the landscape scale whenever there are species present in a landscape that are restricted to engineered habitats. Results from a combination of experimental and observational approaches suggest that species richness within beaver-created wetlands is more strongly controlled by local factors, such as within meadow variability in hydrology, but that the population dynamics of beaver could potentially affect species richness by altering the age distribution of meadows across the landscape. I developed a set of models that links the population dynamics of ecosystem engineers to the dynamics of the patches that they create, and used data on the population dynamics of beaver to estimate model parameters. I used these estimates to perform simulations on the effects of varying parameters on the distribution of different patch types and suggest a number of hypotheses that derive from model predictions that could test the model. The effect of ecosystem engineers on species richness at the patch scale should be predictable based on knowledge of how the engineer affects productivity. Using data collected from the literature, I found a significant negative relationship between productivity and the engineering effect on species richness when ecosystem engineers increased productivity and a significant positive relationship when engineers decreased productivity, matching the predictions of my conceptual model.