Abstract
Previous traditional ecological models of single-species insect populations regulated by weak density dependence did not conform well to the ubiquitous relation between spatial variability and mean density known as Taylor's power law. New evidence is assembled to demonstrate that this may have been because the degree of density dependence modelled was unrealistically small. Similar models with stronger density dependence, which generate simulated populations within the chaotic region, are presented in this paper. Using very few parameters, these models can generate realistic data which conform very well to the power law relation and to observed frequency distributions and time series of British moth species.