Comparison of environmental DNA and bulk DNA metabarcoding for assessing terrestrial arthropod diversity across three habitat types on Guam

DNA based methods offer a rapid and cost-effective way for detecting species occurrence and monitoring biodiversity; among them bulk DNA metabarcoding is well-established, and recently developed environmental DNA (eDNA)-based methods offer a non-destructive alternative. With a goal to develop suitable methods for assessing insect biodiversity in ecosystems for which DNA reference libraries are not well developed and incomplete, such as remote islands, we compared established bulk DNA metabarcoding methods with eDNA across three replicated terrestrial ecosystem types (limestone forest, degraded forest, and grassland) in Guam. Using two mitochondrial COI primer pairs, we performed bulk DNA metabarcoding of standard entomological collection methods (malaise traps, pan traps, vegetation beating), and compared the assessment of biodiversity with that from different eDNA sources (flowers, spider webs, leaves, tree trunks). In our samples, eDNA and bulk DNA metabarcoding both detected a large proportion of overall taxa (OTUs, 86.6% and 60.3%, respectively). Although DNA metabarcoding detected significantly more taxa, eDNA proved to be a reasonable non-destructive alternative. As expected, because of limitations in existing reference databases for remote habitats, species-level identification was achieved for only a few OTUs. Overall, the sampling approach was the dominant driver of arthropod diversity, explaining [~]17% of the observed variation, while habitat type accounted for [~]4%. Thus, each sampling approach captured some unique diversity signals and contributed to the complementary effect of maximizing detection. For rapid insect biodiversity surveys of terrestrial arthropods, we recommend an integrated metabarcoding approach, and in sensitive habitats where insect capture is undesirable, eDNA offers a powerful alternative to monitor diversity and community change.