Structural Stability of Ant-plant Mutualistic Networks Mediated by Extrafloral Nectaries to Forest Fragmentation in the Brazilian Amazon

Rainforest fragmentation drastically affects biodiversity and species composition, mainly due to habitat loss. Several studies have already shown the effects of forest fragmentation on plant and ant communities. To date, however, there is limited empirical knowledge of how forest fragmentation affects ant-plant interaction in networks. We investigated the effects of the configuration of rainforest fragments on the structure of ant-plant interaction networks mediated by extrafloral nectaries (EFNs). We carried out this study in ten forest fragments, ranging in size from approximately 5 to 3,000 ha, located in the Brazilian Amazon. In each fragment we established a plot of 6,250 m2, in which all ant-plant interactions were recorded, and calculated the following network descriptors: number of interactions, network size, network specialization, diversity of interactions, and nestedness. We used four explanatory variables to investigate the effects of forest fragmentation on these network descriptors: three metrics of the configuration of fragments (i.e., fragment area, edge irregularity, and connectivity) and the forest structure within each fragment, represented by canopy cover. We did not detect any effect of the explanatory variables on the network descriptors. The structural stability of the networks sampled in forest fragments with different configurations is possibly related to the observed constancy of ant species in the central core of highly interacting species. Our results corroborate other studies highlighting the structural stability of these facultative ant-plant networks mediated by EFNs in different spatial and temporal gradients. Nonetheless, the low constancy of plant species in the generalist core should be understood as a warning, mainly because the functionality of this protective mutualism (i.e., food secretions in exchange for protection against herbivory) remains unknown.