Aim: To evaluate the extent to which ant species richness in Neotropical savannas varies with macrogeographic variables, and to identify the potential climatic drivers of such variation.
Location: The Cerrado savanna biome of central Brazil, in a region spanning ca. 20° of latitude and 18°of longitude.
Methods: Standardized sampling of the arboreal and ground-dwelling faunas was performed in 29 well-preserved savanna sites using pitfall traps. Species were classified according to their habitat affinities: open-savanna specialists, forest-associated species or habitat generalists. We used generalized linear models to evaluate the importance of geographic (latitude, longitude and elevation) and climatic (mean temperature and three metrics of rainfall) variables as predictors of species richness.
Results: The total number of species recorded at each site varied more than twofold (from 59 to 144), and latitude was the best geographic correlate of overall species richness. However, contrary to the expected pattern, more species were found at higher than lower latitudes. This reversed latitudinal pattern of diversity occurred for both the arboreal and ground-dwelling faunas, and for the habitat generalists and forest specialists. The savanna specialists showed a mid-latitudinal peak in diversity. Overall, there was a significant positive association between rainfall and species richness, but the strength of this relationship varied with ant habitat affinity.
Main conclusions: The Cerrado ant fauna shows a reverse latitudinal gradient in species diversity, and this can be explained by increasing rainfall during the warmest months of the year (and therefore in plant productivity) with increasing latitude. The sensitivity of Cerrado ant diversity to declining rainfall contrasts with the high resilience to aridity of the Australian savanna ant fauna, and this reflects the contrasting evolutionary histories of these faunas. Our findings highlight the importance of historical processes as drivers of intercontinental contrasts in macroecological patterns.