DNA (eDNA) metabarcoding enables broad biodiversity monitoring; however, its accuracy depends on the completeness of the reference databases. Early studies often underestimated or misclassified diversity due to the absence of reference sequences. We revisited an eDNA dataset first analyzed in 2014 (Simmons et al., 2016), which surveyed fish communities in the Muskingum River, a part of the Mississippi River Basin (MRB). Original samples (collected in 2013, amplified with 16S primers, and sequenced on an Illumina MiSeq) were reanalyzed against GenBank Actinopterygii 16S reference libraries, which were updated annually from 2014 to 2024. During this period, the database nearly doubled in size. Species richness estimates remained relatively stable (46–54 taxa), but species identifications improved: non-MRB detections declined while MRB taxa increased, reducing spurious matches. The ratio of non-MRB to MRB detections dropped from 0.2 (2014–2019) to 0.06 (2024). Metrics of overlap, gain, loss, and turnover show that new reference sequences primarily drove species gains, while losses often reflected correction of errant matches. These dynamics highlight how database growth reshapes biodiversity inferences and reinforce the value of archived eDNA as evolving digital records for conservation monitoring.