Estimating the reliability of diet overlap measures

Synopsis Diet overlap measures, commonly used in studies of resource partitioning and competition in fish, are too often treated as fixed values; in fact, they are random variables. Two methods for estimating the variance of some overlap measures using stomach content data are examined here: the jackknife and the bootstrap. Simulation results indicate that the methods work well. In addition, they are valid for either continuous (biomass, percent volume) or discrete (abundance) data. Applications indicate that outliers, such as a stomach much different than others, may have large influence in determing the overlap value.