Environmental DNA (eDNA) techniques are increasingly employed in biodiversity monitoring of terrestrial animals, plants, and fungi, holding great potential to revolutionize biodiversity assessments on land. However, sampling and basic laboratory protocols still require refinement to optimize DNA metabarcoding performance. Homogenization as a pretreatment for eDNA extraction is known to enhance the concentration and quality of extracted eDNA for some groups of organisms. We developed a simple and efficient method for capturing arboreal biodiversity using stemflow as a source of eDNA; however, its performance with or without homogenization had not yet been compared. In this study, we evaluated the performance of two different homogenization methods using eDNA metabarcoding and qPCR assays. Metabarcoding analyses revealed that the method without homogenization detected the fewest species, while nearly identical and higher numbers of species were detected in samples subjected to bead-beating and frozen bead-beating homogenization. Similarly, qPCR analyses revealed that the method without homogenization yielded the lowest DNA concentration, while nearly identical and higher DNA yields were observed for bead-beating and frozen bead-beating homogenization. These findings suggest that, considering cost and effort, the bead-beating method without freezing is the most advantageous.