Bioinformatic analysis of metabarcoding marine data for the development of a Northern Adriatic Digital Twin (NADT)

Within iNEST (Interconnected Nord-Est Innovation Ecosystem), the PNRR-funded innovation ecosystem for Northeastern Italy, Spoke 8 is developing a digital twin of the Northern Adriatic by integrating observations and models. This effort requires biodiversity signals that remain comparable across years, methods, and field constraints. Molecular approaches can scale across complex communities and complement morphology-based surveys, but routine implementation still faces barriers: formalin fixation damages DNA; marker and primer choice forces trade-offs among universality, taxonomic resolution, and performance on degraded templates; reference coverage remains uneven for many zooplankton groups; and field logistics constrain replication and sampling effort. This thesis tackles these barriers through methodological tests in the Gulf of Trieste. It establishes feasibility for metabarcoding on bulk mesozooplankton preserved for seven years in formalin by using a hot alkaline lysis buffer to promote cross-link reversal, followed by proteinase K digestion and amplification with three primer sets. Marker behavior diverged under severe fragmentation: 18SV4 predominantly recovered non-metazoan DNA, often fungi, whereas COI and 18SV9 recovered metazoan taxa more consistently. The bioinformatic workflow uses DADA2 for noise reduction and adds abundance-based greedy clustering (AGC) to stabilize OTU centroids when preservation damage generates irregular variants. Because preservation disrupts quantitative relationships among taxa, the analyses prioritize marker performance and qualitative detection; even under these constraints, the data yield ecologically relevant signals, including detections consistent with non-indigenous species in the regional plankton. The thesis then improves recovery from the same archival material through iterative optimization of extraction and amplification, focusing on the long COI marker under low yield and inhibition-prone conditions. It introduces ddPCR-based amplification for long-term formalin-preserved zooplankton bulk, an approach not previously applied to this type of material, and pairs it with an updated informatic workflow. The optimized protocol reduces cycle demands, supports more conservative filtering, and improves recovery of low-frequency lineages without changing broad phylum patterns. These results remain preliminary, but they show that ddPCR can strengthen COI recovery and data reliability in this preservation context. To strengthen interpretation beyond “what amplifies,” the thesis generates targeted local reference sequences from freshly collected zooplankton and documents primer trade-offs that affect downstream usability. It shows how reference composition can yield ecologically implausible top hits for larval stages and how highly degenerate COI primers can amplify strongly yet fail in Sanger sequencing, which constrains reference building even when PCR succeeds. Finally, within CenSy4Sea (a PNRR National Biodiversity Future Center, NBFC, cascading-grant project), the thesis evaluates passive eDNA filtration as a field-practical complement to active filtration. Of four matrices tested, fiberglass, gauze, and sponge yielded amplifiable DNA. The three effective matrices produced comparable community profiles, while primer choice drove yield and composition, with Teleo2 showing the most consistent performance among the tested 12S primer sets. Passive filtration recovered most catch species and produced a more stable, time-integrated signal than active transect filtration. Overall, this work delivers traceable molecular biodiversity observations and explicit operational limits that help iNEST ingest, compare, and reuse biodiversity data across methods. Under urgent monitoring needs, method development requires iteration; documenting constraints, missteps, and low-yield outcomes as part of the evidence helps many small teams build shared, robust monitoring rather than isolated snapshots.