Land use is a stronger determinant of ecological network complexity than the number of trophic levels

Land modification causes biodiversity loss and ecosystem modification. Despite many studies on the impacts of this factor, there is little empirical evidence on how it affects the interaction networks of plants, herbivores and their natural enemies; likewise, there is little evidence on how those networks change due to differences in the complexity of the communities they comprise. We analyzed the effects of land use and number of trophic levels on the interaction networks of exotic legume species and their associated arthropods. We collected seedpods from five exotic legume species (one of them invasive) in four land use types (urbanization, roadside, L. leucocephala plantation, wooded pasture) on Santa Cruz Island in the Galapagos, and obtained all arthropods that emerged from the seeds. Then, we built and analyzed the interaction networks for each land use at two community scales, each with different numbers of trophic levels: (1) three levels: plant-seed beetle-parasitoid (PSP), and (2) more than three levels: plant-seed beetle-parasitoid-predator and other trophic guilds (PSPP). Land use was more relevant than number of trophic levels in the configuration of species interactions. The number of species and interactions was highest on roadsides at PSPP and lowest in plantations at PSP. We found a significant effect of land use on connectance and interaction evenness (IE), and no significant effect of number of trophic levels on connectance, diversity or IE. The simultaneous analysis of land use and number of trophic levels enabled the identification of more complex patterns of community structure. Comparison of the patterns we found among islands and between exotic and native legumes is recommended. Understanding the structure of the communities analyzed here, as well as the relative contribution of their determinants of change, would allow us to develop conservation plans according to the dynamics of these neo-ecosystems.