Abstract

Environmental DNA (eDNA) metabarcoding is a powerful tool for monitoring aquatic biodiversity in extreme environments like the Qinghai–Tibet Plateau (QTP), where traditional methods face significant challenges. However, key knowledge gaps remain, particularly in selecting optimal sampling substrates for target taxonomic groups and understanding substrate-specific detection differences. Here, using eDNA metabarcoding with five primer sets (targeting mitochondrial, nuclear, and nucleoid sequences), we systematically compared biodiversity detection between sediment and water samples collected from five sites in the Dongjicuona Lake across the tree of life (encompassing archaea, bacteria, algae, fungi, plants, protozoa, invertebrates, fishes, birds, and mammals). The results showed that sediment eDNA exhibited significantly higher amplicon sequence variant (ASV) richness compared to water eDNA for most taxonomic groups. In contrast, the opposite pattern was observed for fishes, while no significant differences were observed for fungi and birds. Community composition differs significantly between sediment and water samples for most taxonomic groups, except for mammals. Across all taxonomic groups, 18.0% of ASVs were shared between sediments and water, with fishes showing the highest overlap (42.6%) and archaea the lowest (4.5%). Additionally, the percentages of ASVs contributing to major community dissimilarity (cumulatively accounting for 80%) ranged from 6.4% (invertebrates) to 17.4% (protozoa), with most taxa exhibiting distinct habitat preferences. These findings support water sampling as an effective method for monitoring fish diversity, while emphasizing the need for a multi-substrate approach to comprehensively monitor other taxonomic groups in plateau lakes.