Introduction
Interactions between species are a key component for understanding biodiversity (Abram 1987). In fact, individuals of all species rely upon ecological interactions to obtain food, to breed, or to protect against natural enemies (Thompson 2005). Ecological interactions form networks that connect populations of different species in a locality (Bascompte and Jordano 2013). The organisation of these networks may have important conservation implications, potentially affecting the robustness of ecological systems to species loss (Schmitz and Beckerman 2007). In this context, it is essential to understand how factors that influence the interactions between individuals affect the structure of networks at the level of ecological communities.
The structure of several ecological networks generally deviates from what is expected for networks in which individuals interact randomly, i.e., the interaction is proportional to the product of species abundances (Krishna et al. 2008). These deviations from expected network structure suggest that factors such as the characteristics of interacting individuals and environmental conditions influence the structural patterns of ecological networks at the community level. Among the traits that may affect network structure is body size, which is directly associated with the ability of individuals to consume resources (Stouffer et al. 2011). If the larger the predator individual, the greater its ability to kill larger prey we should expect, at the species level, larger average body sizes to be correlated with a larger number of resource types consumed by predator species (Sinclair et al. 2003). Furthermore, if only body size were influencing the capacity to consume a wider range of resources, it is expected that the diet of the smaller predator species will be a subset of the items of the larger predator species’s diet, leading to nested ecological networks (Sinclair et al. 2003, Woodward et al. 2005, Stouffer et al. 2011).
On the other hand, food resources are not homogeneously distributed in environments and the degree of specialisation in the consumption of distinct sets of prey may require distinct adaptations (Schoener 1968, Covich and McDowell 1996). For instance, the Anolis lizards of South Bimini islands divide habitat and the food resources according to lizard average size classes, in which larger lizard species usually eat larger food items than smaller lizard species (see Schoener 1968). Thus, we can expect that due to the restrictions related to prey handling, prey detection, or nutritional yield, larger predators are predisposed to disregard smaller prey (Mittelbach 1981, Arnold 1993, Arim et al. 2010). In this sense, deviations from the perfectly nested pattern are expected, enabling the formation of semi-isolated groups (modules) in the network. Networks with a modular structure have stimulated much interest due to its possible evolutionary and ecological consequences (Ings et al. 2009). For instance, modules may represent coevolutionary units (Thompson 2005) and increase the stability of ecological networks, thus providing a potential mechanism through which complexity arise and persist in ecological communities (Krause et al. 2003, Ings et al. 2009).
Here, we explore the trophic network organisation of a community Amazonian snakes. Snakes have been used as a model system in studies on the effect of ecological interactions on diversity (Martins et al. 2001, Colston et al. 2010, Alencar et al. 2013, Bellini et al. 2015, Klaczko et al. 2016, Alencar et al. 2017). These studies explore how ecological traits, interspecific interactions, habitat use, and evolutionary history influence the current trophic interactions of different species of snakes. As snakes evolved morphological and behavioral adaptations to kill and ingest their prey whole (Greene 1983), traits related to different dietary habits of species make snakes a model study system to understand how trophic interactions organise community structure (Shine and Bonnet 2000).
Motivated by understanding the trophic organisation pattern of snake communities, we here use as a model a rich and well-studied community of Amazonian snakes (Martins and Oliveira 1998). We characterised the structure of the interaction network between snakes and their food resources. We expected that if only snake body size were shaping network patterns at the species level, the structure would be nested; on the other hand, if specialization in resource consumption were driving patterns of resource use across snakes, modularity would be expected. We then evaluated the role of different snake species and the effect of snakes’ habitat use (referred to here as a lifestyle) in shaping the network structure.