Assessing and conserving marine biodiversity remain critical global challenges, particularly in highly disturbed coastal regions. The use of environmental DNA (eDNA)-metabarcoding has revolutionized biodiversity assessment and management
however, the prevalence of both false positives and negatives continues to be a significant concern. To address these technical errors, we tested two potential methodological improvements in the highly disturbed Guangdong-Hong Kong-Macao Greater Bay Area: (1) the use of random whole-genome amplification (WGA) to reduce false negatives derived from low eDNA concentration, and (2) the application of environmental RNA (eRNA)-metabarcoding to mitigate false positives arising from eDNA contamination by human activities. Using fish communities as our target, we found that WGA enhanced downstream PCR amplification for metabarcoding but significantly reduced the detection of rare taxa, altered community structure, and increased false negatives (p <
0.002 for all tests). Interestingly, WGA led to higher levels of false negatives in more biodiverse communities. eDNA-metabarcoding revealed that 20.9%-23.6% of detected taxa were pure freshwater species (false positives) incapable of surviving in estuarine and coastal regions, highlighting the often-overlooked eDNA contamination in disturbed coastal ecosystems. In contrast, eRNA-metabarcoding significantly reduced false positives (p <
0.002), with error taxa accounting for only 2.5%-6.3% of all detections. Comparisons between eDNA and eRNA metabarcoding further revealed differences in their rare taxa recovery capacity. The findings provide critical insights into method selection for biodiversity assessment and management in highly disturbed coastal regions and highlight the need for further technical improvement of eDNA and eRNA-based biodiversity monitoring and conservation in aquatic ecosystems.