Combined water and detritus sampling is the most effective strategy to detect environmental DNA of macroinvertebrates in rivers

• Systematic comparison of five metabarcoding sources (water/sediment/detritus/ethanol/bulk) • eDNA samples detected more macroinvertebrate species and OTUs than DNA samples. • Water samples were more effective in detecting bivalves. • Detritus samples were more effective in detecting oligochaets. • Water and detritus sampling enables comprehensive monitoring of macroinvertebrates. Although environmental DNA (eDNA) metabarcoding technology has been increasingly applied in freshwater ecosystems, the effectiveness of conventional water sampling and other adaptive metabarcoding methods for detecting macroinvertebrates has not been comprehensively compared and validated. Water, animal tissues (bulk samples), detritus and sediment are regarded as good materials to detect metabarcoding for macroinvertebrates. In this study, we collected DNA metabarcoding samples (bulk samples, BS) and eDNA metabarcoding samples (water samples, WS; detritus samples, DS; detritus ethanol soaking solution samples, DES; sediment samples, SS) in the mid-lower Han River and comprehensively illustrated the pros and cons of different samples. We compared the composition structures of macroinvertebrate community obtained by the above methods, analyzed the detection efficiency of different sampling combinations, and proposed the most effective strategy. The results showed that more macroinvertebrate species and Operational Taxonomic Units (OTUs) were detected using eDNA metabarcoding compared to DNA metabarcoding. The macroinvertebrate sequences showed that BS (151,841) > DS (87,421) > DES (21,757) > SS (13,037) > WS (8,606). WS showed superior detection of bivalves but failed to detect Plecoptera, Enchytraeida, Lepidoptera and Isopoda. DS detected almost all orders and was the most effective in detecting oligochaetes. The composition structures of macroinvertebrate communities in WS and DS were significantly different ( p = 0.004) and complementary. Our results demonstrated that the combination of WS and DS provides a more comprehensive detection of the composition structure and diversity of macroinvertebrate community. We hope our findings can promote the application and standardization of eDNA metabarcoding technology for monitoring riverine macroinvertebrates.