The potential species richness (PSR) of tropical forests is known to be related to climatic factors. However few studies have mapped the current and future spatial pattern of potential species richness for large numbers of species at a regional scale. Our study looked at PSR of Mesoamerican tree species from Panama to Southern Mexico. The region has strong gradients in seasonality, rainfall and temperature. We asked three questions. Does PSR show a unimodal, mid domain effect? Is the shape of the relationship between PSR and temperature affected by seasonality in rainfall? How are potential species loss, gain and turnover under climate change related to the current spatial pattern of PSR?

We used generalized additive models (GAMs) to predict potential distributions for 2000 tree species using herbaria records and climate layers. Results from general circulation models were downscaled to produce layers that represent the possible climate in 2100. Fitted GAMs were used to predict distributions under climate change, assuming that species are free to disperse. Outputs from the distribution models were combined in order to map spatially explicit indices of potential species richness. Conditional spline models were then used to relate current PSR, future PSR and species turnover to climatic factors. 
Results/Conclusions

Length of the dry season was found to be the most important determinant of PSR at a regional scale. Maximum values for PSR were found at moist, mid elevations with mean maximum monthly temperatures below 24 ºC. A unimodal, humpbacked relationship, between maximum temperature and PSR was apparent in areas with at least nine months in which monthly rainfall exceeds 100 mm. However this mid domain effect was not found in areas with more seasonal rainfall, suggesting that it is not an inevitable result of constraints to species ranges.Species loss under climate change also showed a unimodal pattern. Potential species loss under climate change was concentrated at mid elevations. However the potential for gains in species richness increased monotonically with elevation.

The results support the view that mid elevation tropical montane forests may be under greater threat from climate change than lowland forests.