A temporal snapshot of marine fish eDNA metabarcoding reveals the promise and the limitations of comparative analysis

1. Context and purpose: Environmental DNA (eDNA) is increasingly used for fish biodiversity monitoring in marine ecosystems, yet methodological pipelines remain highly heterogeneous and continue to diversify. Although numerous reviews describe available experimental and bioinformatic approaches, explicit links between methodological choices and species detection outcomes remain scarce. This study therefore assessed recent marine fish eDNA metabarcoding studies to explore associations between methodological choices and species-level detections. 2. Approach and methods: A comparative dataset was compiled from recent marine fish eDNA metabarcoding campaigns, including information on sampling design, filtration, replication, spatial coverage, and other methodological variables. Associations between these methodological choices and reported species-level detections were explored descriptively. Comparisons were strongly constrained by inconsistent reporting and incomplete methodological and taxonomic metadata. 3. Main results: Species-level detections increased with the total number of samples and were higher in campaigns using onsite filtration. In contrast, several approaches commonly regarded as good practice, including larger sampling volumes, greater spatial coverage, technical replication, and transect sampling, were not associated with higher species-level detections in the compiled dataset. An exceptionally successful campaign also suggested that temporal replication may represent an underexplored dimension of sampling effort. 4. Conclusions and wider implications: This analysis identifies a limited number of methodological variables associated with higher species-level detections while revealing a persistent gap between established good practices and their implementation. More importantly, it demonstrates that robust large-scale comparisons are currently limited by the lack of standardized, accessible taxonomic data and methodological metadata. Coordinated, FAIR-compliant reporting and shared data infrastructures are therefore needed to support evidence-based, reproducible, and continuously updated syntheses of eDNA methodological performance.