Although latitudinal gradients in species richness are well-documented for a plethora of taxa in terrestrial, freshwater, and marine environments, little consensus exists concerning the predominant biological factor that is responsible for the pattern. We produced an analytical null model to assess the degree to which gradients in species richness could be a consequence of the random determination of the limits of species ranges. The model predicts a parabolic increase in species richness toward the middle of a latitudinal domain in the absence of underlying environmental gradients. Our stochastic model accounted for a significant portion of variation in marsupial and bat species richness for each of three different latitudinal domains in the New World: continental limits, the latitudinal extent of each higher taxon, and the smallest latitudinal extent which comprises 95% of the species in the higher taxon. A unique prediction of the stochastic model, which distinguishes it from all other hypotheses, is that parabolic latitudinal gradients in richness should exist for species wholly contained within random latitudinal subsets. Observed gradients for New World marsupials and bats document that this is true. Regardless of taxon or domain, differences between observed and expected species richness (residuals) were not related appreciably to latitudinal band area (r(2) < 0.15). The ubiquity and similarity of latitudinal gradients in species richness for different taxa could be a consequence of pervasive stochastic mechanisms rather than a product of a dominant underlying environmental gradient to which all species respond. Application of our null model to other gradients (e.g., depth, productivity, disturbance) may provide insight into mechanisms affecting patterns of species richness in other ecological or biogeographic settings.
