Using DNA metabarcoding to assess beetle biodiversity for ecologically sustainable forest management

How mature forests in landscapes could influence the recovery of biodiversity in harvested areas is important for guiding landscape planning of ecologically sustainable forest management. DNA metabarcoding is a DNA-based species identification method with great potential for characterising these influences. This thesis tested the capacity for DNA metabarcoding to characterise the highly diverse Tasmanian wet forest beetle fauna compared to traditional morphological identification methods. I applied DNA metabarcoding in a landscape ecology study to better understand the main landscape drivers of beetle communities in late-successional regeneration forests. This thesis includes a review and three studies. Chapter 1 is a general introduction that provides background regarding biodiversity conservation in managed forests, highlights the potential and limitations of DNA metabarcoding for biodiversity monitoring, and identifies the research gaps in understanding the importance of retained mature forest at the landscape scale. This chapter also outlines the structure of this thesis. Chapter 2 provides a practical guide for entomological ecologists seeking to apply DNA metabarcoding in their own studies. I review approaches to field sampling invertebrates for DNA preservation, highlight important considerations for laboratory work, and suggest graphical user interface tools for bioinformatic analyses in a standard workflow for DNA metabarcoding. This manuscript is published in Ecological Entomology. Chapter 3 demonstrates that DNA metabarcoding can capture subtle beetle community composition differences, albeit with slight signal loss compared to morphological identification. This was based on a comparison of beetle community composition using presence/absence data of beetles collected with pitfall traps at 12 regeneration forest sites (~55-year-old) and 12 neighbouring unlogged mature forest sites. DNA metabarcoding of the cytochrome c oxidase subunit I gene (COI), but not the mitochondrial 16S ribosomal RNA gene (16S), identified similar number of zero radius operational taxonomic units (176 and 156 ZOTUs for COI and 16S, respectively) compared to the number of species identified by morphological identification (173), suggesting the suitability of mitochondrial COI for studying invertebrate biodiversity. This study also highlighted the incompleteness of DNA barcode reference databases for assigning species, with only 12.5% COI and 16% 16S ZOTUs being assigned to species, respectively. This manuscript is published in Restoration Ecology. Chapter 4 used the dataset from Chapter 3, but restricted analysis to the subset of ZOTUs identified to species level, to study the performance of DNA metabarcoding for estimating species biomass as a proxy for beetle abundances. The results showed a significant correlation between input species biomass and output high-throughput sequencing DNA read abundance for the 16S dataset but not COI dataset. The number of primer-template mismatches was higher in COI (0-7 bp) than in 16S (0-2 bp), affecting the performance of quantitative DNA metabarcoding. However, the type and the total number of COI primer-template mismatches showed strong phylogenetic signal for estimating species biomass. Researchers seeking quantitative results from metabarcoding results should choose genes with few or no primer-template mismatches or improve the accuracy of species biomass estimation with phylogenetic corrections with markers like COI. This manuscript is in preparation for submission. Chapter 5 investigated the relative importance of the proximity to ('forest influence') and the amount of ('landscape context') mature forests in the surrounding landscape for beetle biodiversity recovery in late-successional regeneration forests (40-58 years old). While the beetle community composition did not fully recover to conditions similar to those in mature forests, the landscape context was relatively more important in affecting the overall beetle composition than the forest influence. However, their importance was fairly low in late-successional regeneration forests compared to the spatial position of study locations. Maintaining a mix of regeneration age classes and mature forests in the landscape could help short- and long-term biodiversity conservation in managed forests. This manuscript is published in Forest Ecology and Management. Chapter 6 synthesizes the results of the preceding chapters and discusses the application of DNA metabarcoding to invertebrate biodiversity monitoring, and the implications of the field studies for forest managers. In addition, this chapter suggests some directions for future research. My thesis demonstrates that DNA metabarcoding provides a cost-effective tool for studying invertebrate biodiversity and probing subtle community composition changes. It highlights the importance of comprehensive DNA barcode reference databases from an applied perspective. Moreover, further research is required to enable species biomass or abundance to be accurately estimated with DNA metabarcoding. Both forest influence and landscape context impact beetle biodiversity recovery, although the factors including the fine-scale spatial position of sites and landscape configuration of mature forest in the landscape are important factors structuring beetle assemblages in harvested areas. These findings provide insights relevant to landscape conservation planning in managed forests.