Abstract
Estimates of global species richness for insects are based upon extrapolations from “known” to unknown faunas and hence rely upon accurate counts of species for the referrent taxon or region. The number of reference species depends upon the species concepts employed by workers in that group combined with the degree to which nonstandard (i.e. nonmorphological) approaches have been used. Genetic data are more directly applicable to the detection of the apparent absence of gene flow, which lies at the heart of any species concept, than is morphological information. But what criteria can be used as a practical guide to suggest the absence of gene flow and define species-level units? Minimally, the phylogenetic species concept requires that there be one fixed difference between two samples for them both to be considered discrete species. The assumptions accompanying this definition include the survey of sufficient geographic locations, loci, and individuals. Based upon six studies of mostly widespread, readily identifiable and well-investigated bee “species”, we estimate that the number of species currently recognised may underestimate the true figure by half (although for at least two of the studies localities have been undersampled and more collections are needed). Even when examples for which there are fewer than five fixed differences between samples are removed from the data set, the number of recognised species increases by perhaps as much as 50% (the same caveat regarding undersampling of populations still applies). We suggest that the presence of morphologically unrecognised species may be more common among widespread, easily identified “species” than is generally accepted. Whether or not similar levels of species underestimation apply to other faunas, such as tropical rainforest canopy beetles, remains to be investigated.