On the processes generating latitudinal richness gradients: identifying diagnostic patterns and predictions.

We use a simulation model to examine four of the most common hypotheses for the latitudinal richness gradient and identify patterns that might be diagnostic of those four hypotheses. The hypotheses examined include (1) tropical niche conservatism, or the idea that the tropics are more diverse because a tropical clade origin has allowed more time for diversification in the tropics and has resulted in few species adapted to extra-tropical climates. (2) The ecological limits hypothesis suggests that species richness is limited by the amount of biologically available energy in a region. (3) The speciation rates hypothesis suggests that the latitudinal gradient arises from a gradient in speciation rates. (4) Finally, the tropical stability hypothesis argues that climatic fluctuations and glacial cycles in extratropical regions have led to greater extinction rates and less opportunity for specialization relative to the tropics. We found that tropical niche conservatism can be distinguished from the other three scenarios by phylogenies which are more balanced than expected, no relationship between mean root distance (MRD) and richness across regions, and a homogeneous rate of speciation across clades and through time. The energy gradient, speciation gradient, and disturbance gradient scenarios all produced phylogenies which were more imbalanced than expected, showed a negative relationship between MRD and richness, and diversity-dependence of speciation rate estimates through time. We found that the relationship between speciation rates and latitude could distinguish among these three scenarios, with no relation expected under the ecological limits hypothesis, a negative relationship expected under the speciation rates hypothesis, and a positive relationship expected under the tropical stability hypothesis. We emphasize the importance of considering multiple hypotheses and focusing on diagnostic predictions instead of predictions that are consistent with multiple hypotheses.