The Investigation of Saturation in the Species Richness of Communities: Some Comments on Methodology

Abstract

A central task for ecologists is to explain the variation in local species diversity or richness. Often, this has been undertaken by reference to a limit to community richness that may be set by local ecological factors such as the physical environment, the interactions among species (e.g. competition, predation, parasitism) (Cody and Diamond 1975) and the disturbance regime (Huston 1979). Communities that have reached this upper limit on species richness are said to be 'saturated' (Terborgh and Faaborg 1980). However, the diversity of different communities often fails to converge to a single upper limit under similar conditions (Ricklefs 1987). In these instances, it is necessary to refer to limiting effects caused by biogeographic processes that operate at regional levels, such as the process of colonization from the surrounding region (Cornell 1985a, b, Ricklefs 1987). If the species richness of a locality is limited by the supply of colonising species from the surrounding region, then the community is said to be 'unsaturated'. In order to know whether to invoke primarily local ecological or regional biogeographic explanations, it becomes necessary to discover whether communities are saturated. One way to address this question is to examine the relationship between local species richness (LSR) and regional species richness (RSR) in standardised samples taken from comparable habitats from different geographical areas (Terborgh and Faaborg 1980, Cornell 1985a, b). If local ecological factors have limited the species richness (i.e. the communities are saturated), then LSR will reach an upper limit that is independent of increasing RSR. If, instead, regional processes govern local diversity then, all things being equal, LSR should increase with increasing RSR. Many studies that have used this basic procedure have concluded that there is little evidence for community saturation (e.g. Cornell 1985a, b, 1985, Ricklefs 1987, Cornell and Lawton 1992, Hawkins and Compton 1992, but see Aho 1990, Aho and Bush 1993, Kennedy and Guegan 1994). Here, we suggest a method to remove a possible source of bias in the typical method of analysis and then we make some additional suggestions regarding statistical testing. Consider the definitions of RSR and LSR, which roughly correspond to Whittaker's (1972) definitions of alpha and gamma diversity. We use the following set terminology: