Introduction

A central goal of biodiversity research is to understand processes of species coexistence at different spatial and temporal scales. Much empirical research has revolved around documenting patterns of species abundance and distribution with sound sampling techniques and statistics (May, 1975; Magurran, 1988; Krebs, 1998). Such data are of tremendous importance not only for documenting current biological diversity patterns, but also for testing fundamental ecological theories (Ricklefs & Schluter, 1993; Brown, 1995; Hubbell, 2001). A common feature of these approaches is the emphasis placed on species as the appropriate currency for quantifying biological diversity. However, documenting species diversity often represents a considerable practical challenge. First, no one exactly knows the total number of extant species on Earth (Erwin, 1982; May, 1994; Novotný et al., 2002; Alroy, 2002). Second, in any given sample, a sizeable fraction of the individuals may represent previously undescribed species, as is especially the case for lesser known groups, like plants in tropical forests, insects or protists. Third, species recognition usually relies upon a set of morphological cues which are not always observable. Thus, many individuals within a sample cannot be reliably assigned to previously described species. This is obvious in microbial communities, where different operational taxonomic units (OTUs) can only be distinguished by DNA screening or other molecular methods (e.g. see Suau et al., 1999 for a study of the microbial diversity of the human gut, and Green & Bohannan, this volume). This issue is also serious in macroscopic organisms.