By maintaining a forest-like structure, shaded cocoa plantations contribute to the conservation of ants that usually live in the soil, leaf litter or canopy of tropical forests.Here we synthesize the available information on the diversity and community structure of ants in shaded cocoa plantations in the Atlantic forest region of Brazil, compare ant assemblages in cocoa agroforests with forests and other forms of agriculture, and discuss how these shaded plantations contribute to the conservation of the ants in the Atlantic Forest region. We also discuss ants of economical importance and of special interest, including Camponotus, Dolichoderus, Gnamptogenys, Pachycondyla, Pseudomyrmex and other litter dwelling genera. We discuss the situation of the tramp ant Wasmannia auropunctata in the Bahian cocoa-producing region where it is considered as native, and that of the two cryptobiotic genera Thaumatomyrmex and Typhlomyrmex, as well as that of proven and possible endangered army ant and Ponerini species. A total of 192 ant species from four strata were found in extensive sampling of a cocoa plantation with a relatively simple shade canopy (comprised primarily of Erythrina). Species richness in the cocoa plantations corresponded roughly to that of low diversity native forests, and species composition of cocoa plantations was most similar to native habitats (forest and mangroves) while ant composition in other agricultural habitats was most similar to that of urban areas. Although occurrences of Wasmannia auropunctata were similar in cocoa plantations and forests, abundance of Thaumatomyrmex and Typhlomyrmex, generally thought to be rare ants, was relatively high in cocoa plantations. These results, from cocoa plantations with relatively simple shade, demonstrate the importance of cocoa for ant conservation in the Atlantic forest region of Brazil. It is likely that cocoa plantations with a greater number of vegetation strata and higher tree species richness (such as traditional cabruca plantations) provide even more important habitat for ants generally and for ant species of conservation concern.

Several groups of ants display a reproductive cycle in which two phases of adult activity alternate in synchrony with the brood instars. The brood stimulation hypothesis (Schneirla, 1957) was developed for ecitonine army ants to explain the proximate control of such biphasic cycles. According to it, onsets of cyclic activities are triggered by social stimulations arising from the developing brood, rather than by innate pace-makers inbuilt in adult ants. While it seemed to provide an acceptable explanation, this hypothesis failed to be experimentally demonstrated, in spite of numerous fi eld observations. We used colonies of thelytokous populations of the phasic ant Cerapachys biroi as a model in order to test the brood stimulation theory. Brood removal and substitution experiments allowed us to confi rm, fi rst, that the periodicity of the cycle is not controlled by an endogenous rhythm in adults. Moreover, we could also characterise the infl uence of each brood instar on the activity of adult ants. Although we confi rmed the existence of a brood stimulation involved in the control of the cycle, experiments revealed that it was not performed accordingly to Schneirla's hypothesis. In effect, our study suggests a primacy of larval infl uence: the foraging phase was triggered and sustained by larvae-induced excitement rather than by stimulation from the newly-emerged callows.