Spatial and temporal collections provide important data on the distribution and dispersal of species. Regional-scale monitoring invariably involves hundreds of thousands of samples, the identification of which is costly in both time and money. In this respect, metabarcoding is increasingly seen as a viable alternative to traditional morphological identification, as it eliminates the taxonomic bottleneck previously impeding such work. Here, we assess whether terrestrial arthropods collected from 12 pitfall traps in two farms of a coffee (Coffea arabica L.) growing region of Sao Paulo State, Brazil could differentiate the two locations. We sequenced a portion of the cytochrome oxidase 1 region from minimally processed pools of samples and assessed inter- and intraspecific parameters across the two locations. Our sequencing was sufficient to circumscribe the overall diversity, which was characterized by few dominant taxa, principally small Coleoptera species and Collembola. Thirty-four operational taxonomic units were detected and of these, eight were present in significantly different quantities between the two farms. Analysis of community-wide Beta diversity grouped collections based on farm provenance. Moreover, haplotype-based analyses for a species of Xyleborus beetle showed that there is significant population genetic structuring between the two farms, suggesting limited dispersal. We conclude that metabarcoding can provide important management input and, considering the rapidly declining cost of sequencing, suggest that large-scale monitoring is now feasible and can identify both the taxa present as well as contribute information about genetic diversity of focal species.