Ecologia das interações entre formigas e a mirmecófita Tococa guianensis no Cerrado brasileiro

Abstract

Tococa guianensis (Melastomataceae) is one of the few existing myrmecophytic species in the Brazilian Cerrado, and although it is commonly found in the gallery forests (and eventually in grassland vegetation) of this biome little is known about the interaction between ants and T. guianensis. For this I analyzed three populations of T. guianensis in central Brazil. In Uberlândia, Minas Gerais state, I sampled two fragments of gallery forest, in Jataí, Goiás, one grassland and four continuous gallery forests (all with previous human disturbance), and in Emas National Park, Goiás, two undisturbed, continuous gallery forests. In each sampling site I determined the proportion of colonized plants and the species of associated ant. The latter were classified as obligate or non-obligate species, depending whether the species regularly nested in the plant. The effect of obligate ants on herbivory in the host-plant was determined by comparing herbivory levels between plants with obligate ants and plants without ants or with non-obligate ants. Measurements of leaf shape, water content, toughness and density of trichomes were taken in leaves from plants from two sampling sites in Jataí (one in gallery forest and the other in grassland) and one in Uberlândia. Finally, I compared the defensive characteristics (leaf toughness and trichome density) between plants with obligate ants (control plants) and those whose associated ants were experimentally removed. Of the 238 plants found, 67.2% had obligate ants (of which Allomerus sp. was by far the most common species), 21 % had non-obligate ants, and 11.8% no ants. The proportion of plants colonized by obligate ants varied significantly among sites. While in Jataí and Emas over 75% of the plants had obligate ants, in Uberlândia these ants were not found. Although obligate ants were common in the gallery forest of Jataí, they were absent from adjacent grassland. Herbivory in plants without obligate ants was significantly greater than in plants with obligate ants. Herbivory was significantly greater in plants from Uberlândia than in those from Jataí and Emas, and greater in Jataí than in Emas. The absence of obligate ants may explain the greater herbivory in plants from Uberlândia, while differences between Jataí and Emas may be related to the level of human disturbance in these localities. Plants without obligate ants (from a forest fragment in Uberlândia and from a grassland in Jataí) presented significant differences in various leaf characteristics comparatively to plants with obligate ants (from a forest in Jataí). Young and mature leaves from plants with ants had a more elongated shape (lower width to length ratio), a greater water content, and lower toughness and lower density of trichomes in the abaxial (lower) side than those from plants without ants (both from the forest and grassland site). The density of trichomes on the adaxial (upper) side of leaves was independent of the habitat and the presence of obligate ants. Leaves from plants whose ants were removed showed a greater density of trichomes in their lower side than control plants (with ants), but there were no differences in the remaining traits analyzed (leaf shape, toughness, and trichome density in the upper side of the leaves). Although I was not successful in removing obligate ants completely from experimental plants, the observed increase in trichome density suggest that T. guianensis is able to induce alternative defenses when ants are absent or rarei. The (likely) ability of T. guianensis to induce alternative defenses may help to explain the existence and maintenance of viable populations of this ant-plant in the absence of mutualistic ants.