Distinguishing the roles that different factors, such as sampling effects and habitat heterogeneity, play in generating species-area curves continues to be difficult in many communities. A recent response to this challenge is the proposal of a ‘zoom’ protocol in which species richness and habitat heterogeneity are sampled in successively larger units (transects or quadrats). The utility of this approach requires that there be justifiable, predictable and unambiguous relationships between richness and heterogeneity. Results of computer simulations that I have done to test the predicted relationships demonstrate, however, that the predicted patterns were not always observed and, on occasion, more complex relationships were observed in their place. While the development of such protocols may increase our understanding of species-area curves, they are unlikely ever to pronounce unambiguously on their causes.