The compound topology of a continent-wide interaction network explained by an integrative hypothesis of specialization

Abstract

Is there a prevalent pattern among interaction networks: nestedness or modularity? Must consumers always trade-off generalism for average performance in resource exploitation? These two questions have been addressed in various systems, with contradictory results. A recent integrative hypothesis combines both questions within a common theoretical framework, proposing that ecological specialization is structured by different prevailing processes in smaller and larger network units. This should produce both a compound interaction network, formed by internally nested modules, and a scale-dependence on the relationship between consumer performance and generalism. Here, we confirm both predictions in a large dataset on host-parasite interactions. We show that modules indeed constrain nestedness at the whole network level, and that the relationship between parasite generalism and performance on their hosts changed from negative at large to positive at small scales. Our results shed light on both debates, and provide some clues to their integration and solution.

A new model explaining the origin of different topologies in interaction networks

“…, Felix et al. ). Indeed, in nested‐modular simulated networks, nestedness between species of the same module was always much higher than nestedness between species of different modules.…”

“…First, we tested for nested and modular topologies with the use of free null models—null models that do not conserve the modular structure of the matrix (Felix et al. ). We applied a modified version of the algorithm proposed by Vázquez et al.…”

“…Considering the possibility of a compound topology in our simulated networks, we used the approach proposed by Flores et al (2013) and adapted by Felix et al (2017), in which we separately computed the nestedness between species belonging to the same module and the nestedness between species belonging to different modules. In a network with a compound topology we expect the WNODA between species of the same module (WNODA SM ) to be much higher than the WNODA between species of different modules (WNO-DA DM ).…”

“…Beyond the relationship between performance and generalism of consumers, the IHS predictions may be extended to network topology (Felix et al 2017, Mello et al 2018. In networks that contain narrow phylogenetic or functional subsets of species, nestedness should prevail.…”

Revealing in-block nestedness: Detection and benchmarking

“…Thus, most studies have focused exclusively on either of them. The existence of systems which combine both patterns has been largely overlooked, despite challenging indications in natural [21][22][23][24][25][26] and social ecosystems [9]. As of now, the proper identification of such compound structures lays beyond the capabilities of state-of-the-art techniques.…”