Does species diversity really drive speciation?

It has been hypothesized that total speciation rate may depend on species diversity (Seposky 1978), population sizes (Hubbell 2001), and island area (Losos and Schluter 2000). In a recent paper, Emerson and Kolm (2005) tested the hypothesis that species diversity might promote speciation, by comparing the proportion of endemics in a taxon with the number of species in that taxon, in the Canary Islands and the Hawaiian Islands. They show that the number of species of the taxon is a good predictor of the island endemics in that taxon, and suggest that this result demonstrates the role of species diversity in speciation. We propose an alternative explanation: that the same abiotic variables that control species diversity also influence speciation, and that this phenomenon may render the analysis of Emerson and Kolm circular. Emerson and Kolm performed a stepwise regression of the speciation rate of a taxon, measured as the proportion of endemics of that taxon, against a set of independent variables comprising abiotic factors and the numberofspeciesofthattaxon.However,asweshallsee, if the number of species itself is dependent on the same set of abiotic variables as the speciation rate, then the stepwise regression will select the number of species as the explanatory variable for the speciation rate and lead to the wrong conclusion that the abiotic variables are less important. This statistical artefact happens because the number of species accounts for the compound effect of the abiotic variables. Cadena et al. (2005) point out a particular case of this artefact which happens when both species richness and endemism depend on attributes of islands that influence the average age of resident populations and reduce extinction rates, such as area. Here we show that the statistical artefact can occur more generally for any type of abiotic variable that influences both species richness and speciation or extinction rates.