Estimating the biodiversity of terrestrial invertebrates on a forested island using DNA barcodes and metabarcoding data

Invertebrates are a major component of terrestrial ecosystems, however, estimating their biodiversity is challenging. We compiled an inventory of invertebrate biodiversity along an elevation gradient on the temperate forested island of Hauturu, New Zealand, by DNA barcoding of specimens obtained from leaf litter samples and pitfall traps. We compared the barcodes and biodiversity estimates from this data set with those from a parallel DNA metabarcoding analysis of soil from the same locations, and with pre-existing sequences in reference databases, before exploring the use of combined data sets as a basis for estimating total invertebrate biodiversity. We obtained 1,282 28S and 1,610 COI barcodes from a total of 1,947 invertebrate specimens, which were clustered into 247 (28S) and 366 (COI) OTUs, of which <= 10% were represented in GenBank. Coleoptera were most abundant (730 sequenced specimens), followed by Hymenoptera, Diptera, Lepidoptera, and Amphipoda. The most abundant OTU from both the 28S (153 sequences) and COI (140 sequences) data sets was an undescribed beetle from the family Salpingidae. Based on the occurrences of COI OTUs along the elevation gradient, we estimated there are 1,000 arthropod species (excluding mites) on Hauturu, including 770 insects, of which 344 are beetles. A DNA metabarcoding analysis of soil DNA from the same sites resulted in the identification of similar numbers of OTUs in most invertebrate groups compared with the DNA barcoding, but less than 10% of the DNA barcoding COI OTUs were also detected by the metabarcoding analysis of soil DNA. A mark-recapture analysis based on the overlap between these data sets estimated the presence of approximately 6,800 arthropod species (excluding mites) on the island, including 3,900 insects. Estimates of New Zealand-wide biodiversity for selected arthropod groups based on matching of the COI DNA barcodes with pre-existing reference sequences suggested over 13,200 insect species are present, including 4,000 Coleoptera, 2,200 Diptera, and 2,700 Hymenoptera species, and 1,000 arachnid species (excluding mites). These results confirm that metabarcoding analyses of soil DNA tends to recover different components of terrestrial invertebrate biodiversity compared to traditional invertebrate sampling, but the combined methods provide a novel basis for estimating invertebrate biodiversity.