Rapid and precise characterization of eukaryotic marine microalgae in the coastal waters of Singapore with third generation sequencing

Background and aims Coastal cities are witnessing increased urbanization to accommodate to the growing population, but with a parallel reduction in their food production potential and eutrophication of water bodies. Chapter 2 discusses how urban areas like Singapore are exploring the aquaculture industry to fortify food security measures, while being challenged by increased frequencies of harmful algal blooms (HABs) from eutrophication. HABs cause significant economic losses to the fishing and aquaculture industries. For example, the HAB in 2015 along the Straits of Johor (north of Singapore) led to a loss of S$1.3 million per farmer from overnight kills of wild and farmed fishes. Therefore, regular monitoring of coastal waters along with accurate identification of all bloom causing species is necessary. Newer molecular methods are gaining popularity due to their relatively lower costs and precise results. In this thesis, I present a novel long-read rDNA metabarcoding workflow (A: Long-read rDNA amplicon sequencing, B: Bioinformatics pipeline, C: Database of long-read rDNA references) optimized for rapid and precise characterization of eukaryotic marine microalgae, its potential for algal bloom forecasting/detection and capturing dominant taxa in environmental samples. Major findings Chapter 3 describes the first study characterizing eukaryotic marine microalgae through amplicon sequencing of near full-length rDNA (18S-ITS1-5.8S-ITS2-D5 of 28S) using portable MinIONTM sequencers from Oxford Nanopore Technologies (ONT). The custom bioinformatics pipeline exploits intragenomic divergences of the complete internal transcribed spacer region (ITS: ITS1-5.8S-ITS2) for achieving taxonomic resolution at the species and/or sub-species levels, while reducing sequencing error rates from 5% to 0.06 – 0.24%. This workflow is next validated for its specificity using mock communities (4 and 12 species) that mimic environmental samples and for its sensitivity using dilution series (1 cell/mL, 10 cells/mL, 100 cells/mL and 1000 cells/mL) mimicking ‘pre-bloom’ conditions of high biomass blooms in Chapter 4. While species in the mock communities exhibited both interspecific and intergeneric taxonomic resolutions, species of the dilution series were detected at all concentrations. Chapter 5 describes how this workflow has successfully detected dominant genera (Cyclotella, Chaetoceros, Picochlorum and Gyrodinium) along the Straits of Johor, throughout ‘a 2-year time-series’ environmental samples. Of those detected within the fraction of dominant taxa, Gyrodinium, Chaetoceros and Skeletonema were previously reported as ‘HAB-causing’. All error corrected consensus from this workflow will be used to build a database of long-read rDNA references exclusively for microalgal species and will be hosted on the LASeR (Long Amplicon Sequence Repository) database. Conclusion Singapore aims to meet 30% of the local nutritional needs by 2030 through the “30 by 30” vision. Being equipped with methods to better predict HABs would ensure sustainable aquaculture practices which are being scaled up for this vision. The findings of this thesis highlight the potential of the long-read metabarcoding workflow for both regular environmental monitoring and HAB detection. This workflow could complement the current water quality monitoring of chlorophyll measurements in the island and could enable early detection of HABs even in the most remote locations where expensive laboratory facilities and professional expertise are not readily available such as aquaculture farmsDoctor of Philosoph