A recent explanation of the declining species richness gradient with increasing latitude away from the tropics implicated broad scale habitat variability, an associated range expansion, and a resulting increase in niche breadth. The niche breadth in turn was thought to affect richness by competition and rescue effect. While all three factors appear to be correlated, neither the postulated nor alternative causal mechanisms have been tested. We conduct such a test using a system which has all the attributes of the large scale pattern but which, in contrast to continental scale observations, allows for estimation or control of crucial variables such as taxonomic composition, habitat heterogeneity, habitat variability, exact species distribution, and local richness. Specifically, we test the alternative that the correlation between the geographical range of species and local diversity is a function of differential species survival and link this phenomenon to habitat variability. We use 40 species of aquatic invertebrates inhabiting a landscape of 49 miniature rock pools on the coast of Jamaica. The system we examined exhibits a gradient of increasing richness with decreasing habitat variability, analogous to the broad scale latitudinal pattern. Furthermore, species with broader ecological ranges are also broadly distributed. Superficially, this appears to be in agreement with the older explanations but two facts suggest different causes. First, there is no evidence of a ’rescue effect‘ maintaining high richness in many habitats despite their proximity to species sources. Second, ecologically broad species coexist with habitat specialists without reducing richness in jointly occupied habitats.