Effects of predatory ants within and across ecosystems in bromeliad food webs.

Predation is one of the most fundamental ecological processes affecting biotic communities. Terrestrial predators that live at ecosystem boundaries may alter the diversity of terrestrial organisms, but they may also have cross-ecosystem cascading effects when they feed on organisms with complex life cycles (i.e. organisms that shift from aquatic juvenile stages to terrestrial adult stages) or inhibit female oviposition in the aquatic environment. The predatory ant Odontomachus hastatus establishes its colonies among roots of Vriesea procera, an epiphytic bromeliad species with water-filled tanks that shelters many terrestrial and aquatic organisms. Ants may impact terrestrial communities and deter adult insects from ovipositing in the water of bromeliads via consumptive and non-consumptive effects. Ants do not forage within the aquatic environment; thus, they may be more efficient predators on terrestrial organisms. Therefore, we predict that ants will have stronger effects on terrestrial than aquatic food webs. However, such effects may also be site contingent and depend on the local composition of food webs. To test our hypothesis, we surveyed bromeliads with and without O. hastatus colonies from three different coastal field sites in the Atlantic Forest of southeast Brazil, and quantified the effect of this predatory ant on the composition, density and richness of aquatic and terrestrial metazoans found in these bromeliads. We found that ants changed the composition and reduced the overall density of aquatic and terrestrial metazoans in bromeliad ecosystems. However, effects of ants on species diversity were contingent on site. In general terms, the effects of the ant on aquatic and terrestrial metazoan communities were similar in strength and magnitude. Ants reduced the density of virtually all aquatic functional groups, especially detritivore insects as well as metazoans that reach bromeliads through phoresy on the skin of terrestrial animals (i.e. Ostracoda and Helobdella sp.). Our results suggest that the cross-ecosystem effect of this terrestrial predator on the aquatic metazoans was at least as strong as its within-ecosystem effect on the terrestrial ecosystem, and demonstrates that the same predator can simultaneously initiate cascades in multiple ecosystems.