The Payment for Ecosystem Services (PES) is a mechanism to enhance nature conservation. However, the effectiveness of PES is not well understood despite its growing popularity as a solution to reduce environmental degradation and promote sustainable management practices. We evaluated a PES scheme through biodiversity assessment. We collected water and soil samples from 180 headwater catchments in the Atlantic Forest (rainforest): 30 reference samples from protected areas, 45 from areas without PES intervention, and 105 from PES areas. Environmental DNA metabarcoding was employed to gather data on vertebrates (using markers 12SV5, MiBird, MiMammal, and MiFish) and invertebrate species (using COI-R marker 1). Most vertebrate identifications showed over 99% similarity with the reference database. In contrast, 99% similarity was rare for invertebrates, indicating a need for improvement in genomic libraries. We identified 1,324 vertebrate taxa, including rare, critically endangered, and invasive species. For invertebrates, we found 2,445 taxa, including sensitive and invasive species. Water samples from reference sites had more species of insects (60%), fish (70%), amphibians (130%), reptiles (200%), and mammals (10%) than soil samples, except for birds. We observed greater species richness in PES areas (N=1,033), followed by protected areas (N=747) and non-PES areas (N=670). Protected areas contained a higher number of indicator species. The percentage of shared species with protected areas was higher in PES (35.5%) compared to no-PES areas (27%). In water samples, Ephemeroptera, Plecoptera and Trichoptera (EPT) richness was greatest in protected areas, followed by PES and no-PES areas. The NMDS ordination analysis showed no distinct separation between the conservation status of species found in protected PES and no-PES areas, as explained by the short-term monitoring. In conclusion, long-term e-DNA biodiversity monitoring using stream water samples is a promising tool for assessing PES effectiveness.