Foraging activity of an Amazonian leaf-cutting ant: responses to changes in the availability of woody plants and to previous plant damage.

This study investigates some factors affecting the foraging activity of the leaf-cutting ant Atta laevigata. The study was conducted on an abandoned farm near Manaus, in Brazilian Amazonia, where forest was beginning to regenerate. I determined how temporal changes in the structure of the woody plant community (linked to the regeneration process) affected the leaf harvesting activity of this Neotropical ant. A 0.39-ha plot was established, and woody plants existing and emerging in this plot were identified, tagged, measured and mapped. At intervals of approximately 14 days all marked plants were checked to determine which had been attacked by A. laevigata during that period. Woody plant density doubled over the 18-months of the study. However, the number of plants attacked by A. laevigata, controlling for seasonal differences in ant activity, was independent of the number available for attack. The number of plants attacked was also independent of the number of new plants emerged in the previous 2 months. Apart from changes in total plant density, the relative abundance of most species also changed with time. Data on the composition of the plant community and on the composition of the ants' diet at different time intervals were subject to ordination analysis. The analyses revealed that successional changes in the composition of the woody plant community did significantly affect the composition of the leaf diet of A. laevigata. Variations in the ants' diet correlated most strongly with variations in the abundance of Bellucia imperialis. As this increased in abundance its relative contribution to the ants' diet increased. In addition, I observed a small decline in the diversity of plant resources exploited by the ants. B. imperialis was one of the species most preferred by A. laevigata, suggesting that variations in the diversity of its diet were related to the abundance of preferred species. The chances of an individual plant being attacked was independent of whether or not the plant had been attacked during the previous 5 months. This suggests that plant resistance against ant herbivory is not affected by previous damage. Two other lines of evidence also pointed to this. First, there was no significant difference in the leaf area cut (as measured during assays of leaf preference with A. laevigata) between leaves of plants that had been subject to artificial defoliation 3 mo earlier and those of control plants. Second, the number of attacks by A. laevigata on individuals of a given species usually followed a random distribution, even though a uniform or clumped distribution would have been expected under the hypothesis that damage by leaf-cutting ants affects the plant's resistance to further damage.