INTRODUCTION
Mutualistic interactions often involve multiple, non-related species in a community (Bascompte and Jordano 2007b), potentially affecting the ecology and evolution of entire assemblages. However, mutualistic interactions are not distributed homogeneously across the assemblage of potentially interacting species (Vázquez et al. 2005). Rather, some species may interact with many potential partners available in the community, while others interact with only a few (Bascompte et al. 2003, Fagundes et al. 2017, Plowman et al. 2017, Guimarães 2020). This heterogeneous distribution of interactions across species can be influenced by the traits of the individuals and populations involved in the mutualism and their response to the local biotic and abiotic environment (e.g.Vázquez et al. 2005, Miller 2007, Albrecht et al. 2010, Dáttilo et al. 2013b, Maia et al. 2019). Whenever habitat conditions influence these mutualistic traits, it can indirectly affect the patterns of interaction in mutualistic networks and, consequently, the mechanisms shaping natural communities and determining species responses to environmental conditions (see Tylianakis and Morris 2017). Therefore, understand how interactions are distributed in response to environmental conditions can help us to understand the consequences to species interactions over global climatic changes.
Defensive mutualistic interactions between ants and plants with extrafloral nectaries (EFNs) are an example of mutualistic interaction in which ant and plant traits can be influenced by environmental factors (Kersch & Fonseca 2005, Heil 2008, Pringle et al. 2013). Ant competitiveness determines their visitation patterns to plants with EFNs and their quality as bodyguards (Leal and Peixoto 2017, Leal et al. 2022). Dominant ants display a series of aggressive behaviors that ensures them a highly competitive ability (Stuble et al. 2017). This aggressiveness also makes dominant ants better mutualistic partners for plants with EFNs because they are more likely to behave aggressively towards herbivores, unlike subordinate ant species who do not typically exhibit aggression (Buckley and Gullan 1991, Xu and Chen 2010, Flores-Flores et al. 2018, but see Melati & Leal 2018). Like the ants, plant species also vary in their attractiveness to ant bodyguards, with those producing more concentrated and abundant nectar being more appealing to their partners (Blüthgen and Fiedler 2004a, c, Flores-Flores et al. 2018). Therefore, dominant ant species are likely to monopolize the most attractive plant species in the community, displacing subordinate species to less attractive plants (Blüthgen and Fiedler 2004c). At a broad scale, this kind of assortative pairing can predictably shape the patterns of ant-plant interaction along spatial gradients (Dáttilo et al. 2013b), especially along those in which the competitive pressure among ant species is also variable (Leal and Peixoto 2017, da Silva et al. 2019, Lasmar et al. 2021). In this case, ant-plant interactions should become less generalized and form groups of interacting species as the stronger competitive pressure among ant species, with dominant and subordinate ants interacting with a more dissimilar group of plants in the community.
A meta-analysis found that, at a macroecological scale, plants with EFNs benefit more from ant attendance as the environment dries out, due to an increased probability of plant attendance by dominant ant species (Leal and Peixoto 2017). This is because the extrafloral nectar, a water source for ants (Heil 2011, 2015), becomes more valuable to ants when water is scarce, making plants with EFNs more attractive to ants (Ruffner and Clark 1986, Contreras et al. 2013, Leal and Peixoto 2017). However, it neglects that, even in drier habitats, plant species would not be equally valuable to ants. Then, dominant ant species might increase the monopolization of the best plant partners available in drier habitats, increasing the competitive exclusion of subordinate ants from the more valuable plant species. Consequently, it is possible that at a macroecological scale ant-plant interactions become more specialized and grouped as scarcer the water availability – something that can have relevant implications for the eco-evolutionary dynamic of this mutualism at a broad scale. For instance, increasing the contribution of direct effects on trait matching over indirect effects on interactions, accelerating the coevolutionary dynamics on interacting species with the group of interactors.
Here, we investigated how emergent patterns of interactions between ants and plants with EFNs (ant-plant interactions hereafter) at the community level vary over environmental water gradients and evaluated the role of ant competition for plant species secreting extrafloral nectar on the emergence of such patterns. For that, we hypothesized that (i) ant-plant interactions become more specialized and forming sub-groups of interaction partners as drier the environment and that (ii) these effects will be driven by a decline in the overlap of dominant and subordinate ant species in the use of these plants over the environmental water gradients. To evaluate our hypotheses, we used two different approaches based on the ecological network theory (Bascompte et al. 2003, Jordano et al. 2003, Bascompte and Jordano 2007b). In the first approach, we evaluated how the structure of ant-plant ecological networks and plant usage by all ant species varied along a macroecological gradient of water availability (first hypothesis). In the second approach, we evaluated the water availability effect on the patterns of plant usage specifically by dominant and subordinate ant species (second hypothesis).
2. METHODS