Experimental Removal of Insectivores from Rain Forest Canopy: Direct and Indirect Effects

Dial, Roman, and Jonathan Roughgarden. 1995. Experimental Removal of Insectivores from Rain Forest Canopy: Direct and Indirect Effects. Ecology 76:1821–1834

This study considered the effects of insectivorous Anolis lizards on a large, complex food web of arthropods and associated herbivory in a tropical rain forest canopy. We excluded Anolis lizards for 6 mo from 20—30 m high tree crowns in Puerto Rican rain forest. Simultaneous with lizard exclusion, we sampled orb spiders, airborne arthropods, and leaf arthropods in lizard removal crowns and in controls. We also sampled herbivory at the end of the experiment. Lizard removal had strong, statistically significant, positive effects on arthropods >2 mm in length and weak negative effects on arthropods <2 mm. Parameters of arthropod body size distributions differed between removals and controls for leaf arthropods, but not for airborne arthropods. Among arthropod taxa >2 mm, both predatory, i.e., orb spiders and parasitic Hymenoptera, and nonpredatory forms, i.e. Diptera, Coleoptera, Orthoptera, and Blattaria, showed strong significant and positive responses to lizard removal. Large Psocoptera, Homoptera, leaf spiders, and ants did not show significant overall responses to lizard removal. Frequency of herbivore damage on new leaves was positively correlated with abundance of Orthoptera and Blattaria. This damage was significantly greater in lizard removal crowns than in controls, indicating an indirect effect of anoles on plants. The indirect effect of lizards on small arthropods through the predatory anthropod pathway appeared weak. Results of lizard removal shown by this study corroborate other lizard removal studies from more xeric, ground—level habitats with simpler food webs in the West Indies, particularly with respect to orb spiders and herbivory. Taken together with the results of similar experiments performed in trophically less complex systems, this experiment suggests that food web size is less important than body size in determining interaction strength between community members.