The Relationship Between Local and Regional Diversity

The extent to which species richness in local communities is determined by regional and historical processes is not well understood. An increasingly popular way to investigate these large-scale processes is through regressions of local on regional species richness. We sampled local and regional species richness in a broad array of taxa from around the world to address five questions. First, is the relationship between local and regional species richness linear, or does local richness accumulate more slowly at progressively higher regional diversities, suggesting local saturation of species diversity? Second, do these relationships vary with locality size? Third, do taxa and continents differ in the form of relationships between local and regional diversity? Fourth, do relationships between local and regional diversity depart from that expected from a null model in which all individuals of a locality are randomly sampled from a regional pool of species whose abundances have a canonical log-normal distribution? Fifth, using this same null model, how does the expected relationship between local and regional species richness depend on the sampling intensity within localities? We used distribution maps to ensure that diversity was sampled in a consistent manner across diverse taxa. Each region was 500 × 500 km, and localities were 1% and 10% of the region size. There was no evidence of local species saturation, as local species richness was strongly and linearly related to regional richness at both spatial scales. Between scales, local diversity accumulated faster as a function of regional diversity at the larger spatial scale. The slope of this relationship between local and regional diversity was the same among taxa across continents, and between Australia and North America across taxa. In other words, at each spatial scale one relationship between local and regional diversity describes most cases very well. The null model showed that approximately linear relationships between local and regional diversity are expected when regional species abundances are log-normal and when the number of individuals sampled within localities is large (roughly 200 times the number of species in the most species-rich region examined). However, empirical slopes were less than expected from the null model, which we interpret as an effect of spatial turnover of species (beta diversity). Since these slopes were nevertheless similar among taxa and between regions, rates of spatial turnover must be approximately the same among these taxa and regions. The log-normal model also showed that nonlinear (concave down) relationships between local and regional diversity are expected under random sampling when sample size is small relative to regional diversity. Therefore, nonlinear relationships are not necessarily indicative of saturation. Our results suggest that at the scales investigated here local communities are unsaturated and that their diversities are strongly limited by species richness of the surrounding regions. Similarity between taxa and continents in the form of the local-regional diversity relationship implies that “rules” governing the assembly of local communities may be widely consistent. If so, understanding species diversity in local assemblages will require knowledge of processes acting at larger spatial scales, including determinants of regional species richness and spatial turnover of species.