Tree heterogeneity, resource availability, and larger scale processes regulating arboreal ant species richness

Processes acting on different spatial and temporal scales may influence local species richness. Ant communities are usually described as interactive and therefore determined by local processes. In this paper we tested two hypotheses linked to the question of why there is local variation in arboreal ant species richness in the Brazilian savanna ('cerrado'). The hypotheses are: (i) there is a positive relationship between ant species richness and tree species richness, used as a surrogate of heterogeneity; and (ii) there is a positive relationship between ant species richness and tree density, used as a surrogate of resource availability. Arboreal ants were sampled in two cerrado sites in Brazil using baited pitfall traps and manual sampling, in quadrats of 20 m x 50 m. Ant species richness in each quadrat was used as the response variable in regression tests, using tree species richness and tree density as explanatory variables. Ant species richness responded positively to tree species richness and density. Sampling site also influenced ant species richness, and the relationship between tree density and tree species richness was also positive and significant. Tree species richness may have influenced ant species richness through three processes: (i) increasing the variety of resources and allowing the existence of a higher number of specialist species; (ii) increasing the amount of resources to generalist species; and (iii) some other unmeasured factor may have influenced both ant and tree species richness. Tree density may also have influenced ant species richness through three processes: (i) increasing the amount of resources and allowing a higher ant species richness; (ii) changing habitat conditions and dominance hierarchies in ant communities; and (iii) increasing the area and causing a species-area pattern. Processes acting on larger scales, such as disturbance, altitude and evolutionary histories, as well as sampling effect may have caused the difference between sites.