Arthropods serve essential roles in crop production as pollinators, predators, and pests.  Understanding arthropod biodiversity is crucial for assessing agroecosystem health, functions, and services.  Traditional survey methods are labor-intensive, costly, and rely on diminishing taxonomic expertise, limiting their agricultural applications.  Environmental DNA (eDNA) metabarcoding of diverse samples provides comprehensive species composition data through efficient and non-invasive sampling.  However, this method remains underutilized in rice field studies.  This research examined four sample substrates - rice plant cleaning fluid (RPCF), rice pollen, soil, and water - using various barcoding primers to identify optimal substrates for monitoring rice paddy arthropod diversity.  The method was implemented in Bt rice and non-Bt rice fields to evaluate its biomonitoring potential.  Results indicate that the COI primer (mlCOIintF/jgHCO2198R) identified the highest number of rice field arthropod species.  The eDNA collected from RPCF detected 15% more arthropod species compared to vacuum sampling of whole arthropods.  Rice pollen collection during the heading stage also revealed considerable arthropod diversity.  Alpha diversity and taxonomic composition remained consistent between Bt and non-Bt rice fields, aligning with traditional survey findings.  These results suggest that eDNA metabarcoding of plant cleaning fluid offers an effective approach for monitoring agricultural arthropod communities, contributing to agricultural production optimization.