Aridity, but not disturbance, reduces the specialization and modularity of plant–insect herbivore interaction networks in Caatinga dry forest

Plant–herbivore networks comprise over 40% of the global biodiversity and are negatively impacted by habitat loss and fragmentation. However, the extent to which these networks are affected by chronic anthropogenic disturbances and aridity, the most common threats to biodiversity in dry forests, remain unknown. In this study, we examined plant and insect herbivore communities and their interaction networks facing chronic disturbance and aridity gradients in the Caatinga dry forest of northeast Brazil. Our results indicate that aridity and disturbance differentially influenced plant and insect communities. Aridity was accompanied by a reduction in the richness of plant, but not insect species. In contrast, while disturbance was associated with an increase in species richness of sap-sucking insects, it remained unrelated to plant richness. Both aridity and disturbance were identified as factors shaping plant community composition, while insect assemblages only corresponded to aridity. Aridity, but not disturbance, was consistently accompanied by a reduction in network metrics. More precisely, aridity was negatively correlated with network specialization and with the modularity of local webs. Our results also showed that aridity can alter the specialization of specific insect species, suggesting that not only species composition changes can be a key factor modifying network topology, but also shifts in insect behavior across environmental gradients. Our key takeaways are: (1) aridity adversely impacts the dry forest plant community; (2) disturbance can benefit insect feeding guilds such as sap-sucking; and (3) aridity can provoke less specialized interactions at both, species, and network levels. Implications for insect conservation Our results show that aridity disrupts plant–insect herbivore interactions and may trigger cascading extinctions of insect species by reducing the specialization level of their interaction networks.