DNA meta-barcoding and molecular ecology in a forested island ecosystem

Abstract

The development of high-throughout DNA sequencing technologies has resulted in new opportunities for ecological research, including the use of DNA meta-barcoding (hereafter referred to as meta-barcoding) and metagenomic techniques to analyse biodiversity across multiple taxonomic lineages simultaneously. These techniques have great potential to address basic and applied ecological questions based on much broader samples of biodiversity than can feasibly be detected by traditional ecological methods. Soils and related substrates are key targets for DNA meta-barcoding-based biodiversity analyses due to their ecological importance and complex biological composition, including diverse communities of prokaryotes, fungi, protists and invertebrates. A number of methodological uncertainties and challenges currently limit the effectiveness of meta-barcoding techniques for terrestrial biodiversity analyses. These include limited taxonomic coverage of reference DNA sequence databases; the effects of PCR and sequencing artifacts on biodiversity estimates; and uncertainty about the impacts of sampling strategies and DNA extraction methods for analysing eukaryotes (especially animals) in soil communities. This thesis includes three research chapters. The first of these (Chapter 2), describes the compilation of a reference set of DNA barcode sequences from 2000 terrestrial invertebrates collected on Hauturu (Little Barrier Island). These barcodes represent a significant contribution to database coverage of New Zealand invertebrates, with 10 % already represented in Genbank. Taxonomic composition biases were observed between two different invertebrate collection methods, and between the proportions of specimens for which COI and 28S sequences were successfully obtained. Grouping of the barcodes into OTUs provided a basis for investigation of ecological patterns, and comparisons with the results of a parallel 454 pyrosequencing-based DNA meta-barcoding study of soil biodiversity. This indicated that each method tended to recover different components of invertebrate biodiversity, due to the use of contrasting sampling strategies in each case. The total biodiversity of terrestrial invertebrate groups on Hauturu and New Zealand-wide was estimated using a mark-recapture strategy, the results of which were broadly consistent with previously published estimates. Chapter 3 describes an evaluation of methodological factors for DNA meta-barcoding of soil biodiversity, including tests of different DNA extraction methods and an examination of PCR variability. This showed that any of four tested DNA extraction techniques can be used to discriminate spatially separated sample sites at two different scales, based on different components of soil biodiversity. Each method had a bias in terms of the OTUs detected, with less than half of non-singleton OTUs occurring in the results of all four extraction methods, suggesting that the same method(s) should be used if different samples or communities are to be compared. Larger soil DNA extractions resulted in the detection of higher alpha diversity of arthropods (but not micro-organisms) than smaller extracts, and resulted in improved spatial discrimination of sample sites based on metazoan OTUs. The most abundant OTUs were very consistently amplified in replicate PCRs from the same DNA extract, but there was considerable variability of less abundant OTUs, and between different DNA extracts. Species accumulation curves suggested that pooling of ten or more PCR replicates would maximise the number of non-singleton OTUs detected. Recommendations for future soil meta-barcoding studies are provided based on these findings. In Chapter 4 soil DNA meta-barcoding techniques were applied to an analysis of terrestrial biodiversity on Hauturu. Six barcode genes targeting different components of biodiversity were amplified from paired soil samples from 28 plots across the island. The biodiversity of different taxonomic groups was assessed in each sample plot, and hypothetical conservation priorities were identified based on this information. The environmental drivers of biodiversity patterns were investigated, as was the extent to which different genes and taxonomic groups showed the same patterns of community structure. Evidence of elevation-related changes in soil community composition was observed for most genes and taxonomic subsets, most clearly for animal and fungal communities, whereas bacterial and protist communities were more strongly affected by soil pH. Moderately to highly concordant ecological patterns were observed between most genes and taxonomic groups, while fungal community structure was most congruent with above-ground vegetation community structure. These findings are consistent with previous research based on both DNA meta-barcoding and traditional methods. Network analysis methods were used to infer putative ecological interactions among diverse soil organisms and above-ground vegetation species. This resulted in a highly complex network structure, which was divided into 17 communities of closely-linked organisms with contrasting composition. Suggestions for further analysis of this dataset are provided. These results confirm that DNA meta-barcoding is a powerful method for analysing soil communities, with the potential to generate new insights into biodiversity and ecological patterns for diverse organisms in soil, including animals. Further work is needed to maximise the potential of these techniques, however. This includes further consideration of optimal sampling strategies; identification of suitable DNA markers for meta-barcoding of plants; continuing improvements of sequence databases and identification methods; and the development of reliable methods for determination of abundance, although this may necessitate the adoption of metagenomic approaches in the place of PCR-based meta-barcoding protocols. Promising applications for these methods include investigations of small eukaryote biogeography, conservation planning and efficient monitoring of terrestrial environments based on comprehensive samples of biodiversity.