Parasitism is an association based on host individual traits and environmental factors. The complexity of this type of interaction is often lost when studying species-by-species interaction networks. Here we analyze changes in modularity - a metric describing groups of nodes interacting much more frequently among themselves than they do with nodes of other modules, considering the host individual variation and the different forms of parasitism: ecto- and endo-parasitism. For this, we studied mixed networks: bipartite networks comprising host individuals and parasite species as two sets of nodes interacting with each other. We used a fish-parasite mixed network from a highly perturbed coastal river to understand how an anthropogenic perturbation gradient influences the modular structure of host-parasite networks. In addition, we tested how host individual traits drove module configuration within host-parasite mixed networks. Our results showed that different forms of parasitism respond differently to the environment: modularity in fish-ectoparasite networks increased with human perturbation, but modularity was not related to human perturbation in fish-endoparasite networks. In addition, mixed network modules were intrinsically related to individual variation, with host intensity of infection being the most important trait, regardless of the parasite’s life form. The effect of total abundance over network structure indicates signs of changes in community equilibrium, with an increase in species with opportunistic behaviors. Module composition was also related to host fitness and body size, which were most predictive in more preserved and diverse river sections. Overall, our results indicate that host-parasite networks are sensitive to ecological gradients marked by human perturbation and that host individual fitness helps to determine network structure.