Extrapolative nonparametric estimators of species density are commonly used in community ecology. However, they are dependent on either (1) their use on non-dispersive taxa, or (2) the ability to separate tourists from residents in dispersive taxa. We undertook ten years of leaf litter sampling in an ancient woodland in the New Forest, Southern England. We identified all the beetles from those samples and assigned them a residency status (residents, stratum tourists, and habitat tourists). Extrapolations, using the Chao 2, first- and second-order jackknife, and bootstrap approaches, of all sampled beetles all showed large overestimates of species richness when compared with extrapolations based on just residents. We recommend that the estimators should be used with caution as estimates of actual species density for dispersive taxa unless the natural history of most species in a community is well known. This applies especially to tropical ecosystems where many species have not been described. This reinforces the need for more descriptive natural history.