Predator-prey interactions provide key information on the role of each species in the community and an overall assessment of the stability of food webs. DNA metabarcoding has the potential to provide highly informative data which substantially enhance trophic interactions analysis, by providing higher taxonomic detail compared to earlier methods. Here we show, using the Cabo Verde seabird community, that the integrated analysis of trophic networks based on DNA metabarcoding of faecal samples can increase considerably our understanding of the trophic interactions in whole communities. Results revealed that these seabird species prey mostly on fish, with most seabirds relying heavily on very few prey species, which are also targeted by fisheries. This community shows high specialization and modularity levels, i.e., is dominated by seabird species with specialized diets. Such network structure has implications for its management and conservation because specialist predators are especially vulnerable to prey depletion. The Cape Verde shearwater (Calonectris edwardsii), identified as the main network connector species, was confirmed to be a suitable sentinel species of changes in this marine food web. Our results clearly show that network analysis can be used effectively to maximize the potential of DNA metabarcoding in studying trophic interactions of complex communities.