Nonbiological Gradients in Species Richness and a Spurious Rapoport Effect

G. C. Stevens showed that an equatorward increase in species richness is often paralleled by a decrease in the mean latitudinal range of species-a pattern he called Rapoport's rule. He reported a similar pattern for elevational gradients and suggested that both latitudinal and elevational species richness gradients may be a result of the Rapoport effect. Using null models that assume no environmental gradients, we show that "nonbiological" gradients in species richness arise inevitably from the assumption of a random latitudinal (or elevational) association between the size and placement of species' ranges. These models predict a peak in species richness at tropical latitudes and, at intermediate elevations (regardless of latitude), common patterns in empirical data. Using simulated sampling from a parametric distribution of ranges that incorporates a richness gradient but no Rapoport effect, we then show that a spurious Rapoport effect can be caused by sampling bias alone, even when total sampling effort is equal at all latitudes or elevations. The bias arises because per-species sampling effort declines as richness increases, and range is correlated with sample size. Evidence suggests that this sampling bias, which is exacerbated by poorer knowledge of diverse tropical communities, may have affected some of Stevens's elevational results.