Predators, Prey Refuges, and the Spatial Scaling of Density-Dependent Prey Mortality

We tested the biological cause of density-dependent mortality in the bridled goby (Coryphopterus glaucofraenum), a small coral reef fish, and evaluated whether this knowledge allowed us to detect density dependence at different spatial scales in natural habitats. To identify the biological cause of density dependence, we manipulated both population density and the availability of shelter (crevices used as refuges from predators) in small plots of continuous reef. We detected strong density-dependent mortality in plots with few refuges, but mortality was density independent in plots with abundant refuges, indicating that limited shelter causes density dependence. Predator density was unrelated to the density of gobies and refuges, suggesting that predators displayed a type III functional response in patches with few refuges. In a second experiment, we manipulated goby density within replicate plots of three sizes (4, 16, and 64 m2) that varied naturally in the availability of refuges. If refuge availability was ignored, mortality appeared to be density independent at all scales. If, however, plots were grouped by refuge availability, mortality was density dependent in plots with few refuges, but low and density independent in plots with many refuges at all spatial scales. Understanding the mechanism of density dependence (refuge shortage) was thus required to measure the strength of density dependence in natural, spatially variable, habitat. We suggest that density dependence was detectable in plots of different sizes because the relationships between the densities of gobies, refuges, and goby predators were similar across the spatial scales we studied. Our work demonstrates that identifying the biological interactions that cause density dependence, and characterizing the spatial domains at which those interactions operate, will be important to accurately assess the effects of density dependence on population dynamics