SEARCH

SEARCH BY CITATION

Keywords:

  • Boosted regression trees;
  • climatic variability hypothesis;
  • freshwater fish;
  • historical effects;
  • rainfall;
  • physiological tolerance hypothesis;
  • spatial gradients;
  • species–energy hypothesis;
  • temperature

ABSTRACT

Aim

To investigate the effect of climatic, historical and spatial variables on species richness patterns in freshwater fish.

Location

North America and Europe.

Methods

Regional species lists were used to document the spatial richness patterns. Three realms, Europe and Pacific and Atlantic North America, were identified. The numbers of species, by habitat, migration and distributional range categories, were calculated and the contributions of regional mean and seasonal temperature and rainfall, historical (realm, glaciation), and spatial (area, elevational range) variables to predicting richness were assessed using boosted regression trees, model-averaging and spatially explicit models.

Results

The latitudinal temperature gradient is stronger than that for rainfall in the Atlantic realm whereas the rainfall gradient in Europe is independent of the temperature gradient. Species richness is more strongly correlated with temperature than rainfall, and the effects are stronger in the Atlantic realm than in Europe. The influence of environmental variables differs between habitat specialist and generalist species. Climate, particularly maximum monthly temperature, is the best predictor of richness in rivers whereas climate variables are less important than historical/spatial variables for diadromous species.

Main conclusions

Freshwater fish richness differences between realms follow differences in spatial climatic trends. The contributions of climatic, historical and spatial predictor variables vary with ecology: temperature is a better predictor than rainfall in river-dwellers. The richness gradient is driven more by physiological than by energetic constraints on species. The importance of history is probably underestimated because of correlations with climate variables.