Background/Question/Methods   In a simplistic view of a species' range, the edge of the population is both static and coincident with an environmental transition from suitable to unsuitable habitat. Statistical models that correlate the spatial distribution of a species with environmental variables such as temperature and rainfall often imply this view. However, range limits are likely the result of a dynamic interplay between the environment, demography, dispersal, and natural selection. This alternative view suggests that the spatial distribution of the population will not necessarily coincide with the environment and that the location of the edge is dynamic and variable. To test this hypothesis, we examined the relationship between an abiotic environmental gradient and the edge of flour beetle (Tribolium castaneum) populations in highly replicated experimental landscapes. Within these landscapes, we imposed a sharp environmental gradient between wheat flour in which isolated populations maintain positive growth and corn flour in which isolated populations experience negative growth.

Results/Conclusions   Our experimental data reveal three important results. First, the environmental gradient determining the range limit of a species is spatially distinct from the edge of the population. We observed a central (“core”) component of the population in the favorable habitat and a peripheral component of the population extending well beyond the gradient to establish a variable population edge. Second, the population does not experience a sharp transition across the environmental gradient, but instead follows a smooth distribution from the population core, across the gradient, to the edge of the population. Third, variability ultimately determines the distance between the population edge and the environmental gradient. Individuals dispersing from the core are limited in how far they can disperse into the unfavorable habitat and therefore, limited in where they establish the population boundary. However, the severity of that limitation is determined endogenously by demography and dispersal which is intrinsically stochastic and, thus, produces a population boundary at variable distances from the environmental gradient. From these results we can conclude that inferring the quality of underlying habitat from the location of a species range edge is problematic.