Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Drivers of viral diversity and community compositional change over spatial and temporal scales in coastal… Gustavsen, Julia Anne 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2016_september_gustavsen_julia.pdf [ 28.69MB ]
Metadata
JSON: 24-1.0300438.json
JSON-LD: 24-1.0300438-ld.json
RDF/XML (Pretty): 24-1.0300438-rdf.xml
RDF/JSON: 24-1.0300438-rdf.json
Turtle: 24-1.0300438-turtle.txt
N-Triples: 24-1.0300438-rdf-ntriples.txt
Original Record: 24-1.0300438-source.json
Full Text
24-1.0300438-fulltext.txt
Citation
24-1.0300438.ris

Full Text

Drivers of viral diversity andcommunity compositional changeover spatial and temporal scalesin coastal British ColumbiabyJulia Anne GustavsenB.A., English, University of New Brunswick, 2005B.Sc. Hons, Biology, University of New Brunswick, 2005A thesis submitted in partial fulllment ofthe requirements for the degree ofDoctor of Philosophyine Faculty of Graduate and Postdoctoral Studies(Oceanography)e University of British Columbia(Vancouver)April 2016© Julia Anne Gustavsen, 2016AbstractMarine viruses are ubiquitous, abundant, and genetically diverse in natural waters.eyplay key roles in nutrient and carbon cycles. e composition of marine viral commu-nities changes seasonally and repeats annually, and such patterns can be driven by theirhosts in response to environmental changes. Moreover, environmental parameters canalso directly ašect the viral community through the decay of viruses, and dišerencesin viral infectivity under dišerent conditions. Marine viral communities show changesover time and space, but themechanisms that drive compositional changes andmaintainhigh diversity are largely unexplored. Determining factors ašecting viral communitycomposition and structure is essential to explain how viral diversity is maintained.isdissertation will assess the diversity of marine viral communities, and the role of theenvironment and putative viral hosts in driving this diversity.e relationship between environmental parameters and the diversity of viruses andtheir putative hosts was explored in coastal seawater samples along a transect and over a13-month time series at a nearshore location. I usedPCRamplication to target ecologically-important double-strandedDNA(T4-likemyoviruses) and single-strandedRNA(picorna-like) viruses, as well as their putative bacterial (16S rRNA gene) and eukaryotic (18SrRNA gene) hosts were examined. ese were interpreted in the context of nutrients,salinity, and temperature.I observed patchiness in the distribution and diversity of viral communities acrossspace and time (Chapter 2). Chapter 2 greatly increased the known genetic diversityof marine picorna-like viruses with 145 operational taxonomic units (OTUs) occurringwithin previously seen phylogenetic clades. In Chapter 3 there were temporal shiŸs indominance of phylogenetically-related viruses andmost viral OTUs were ephemeral. InChapter 4, I demonstrated that nutrients, salinity, and temperature drive the co-occurrenceiiof viruses and their putative hosts. Finally, in Chapter 5, I revealed that specic viral andprotistan taxa were associated with controlling species composition and the demise of aphytoplankton bloom.Altogether, this dissertation advances the understanding of the phylogenetic struc-ture of viral communities over time, the drivers of host-virus relationships, and the dy-namics of viral and microbial communities during blooms by assessing multiple groupsof viruses and microbes.iiiPrefaceOne chapter from my thesis has been published elsewhere:Chapter 2 has been previously published as:Gustavsen, Julia Anne, Danielle MWinget, Xi Tian, and Curtis A Suttle."High Temporal and Spatial Diversity in Marine RNA Viruses Impliesatey Have an Important Role in Mortality and Structuring PlanktonCommunities." Frontiers in Microbiology 5, no. 703 (2014).doi:10.3389/fmicb.2014.00703.I was the lead investigator on this project. Danielle Winget and Xi Tian, and I had manydiscussions related to the analysis of these data. Xi Tianwrote analytical scripts that wereused to analyse the data. I performed all of the lab work, phylogeny and generated all ofthe gures. Curtis Suttle and I conceived the experiments and we wrote the manuscriptfor the published paper with the input of the other co-authors.All other chapters represent original, unpublished, independent work by the author,J. A. Gustavsen, with the supervision of Curtis Suttle.ivTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . vList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . ixList of Acronyms. . . . . . . . . . . . . . . . . . . . . . . . . . xivGlossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . .xxiii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Synopsis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Microbial inžuence on overall food web . . . . . . . . . . . . . . . . . . . 31.3 Why study temporal dynamics? . . . . . . . . . . . . . . . . . . . . . . . . 41.4 Diversity over time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.5 Community structure of microbial communities . . . . . . . . . . . . . . 71.6 Relatedness over time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111.7 Host-virus interactions over time . . . . . . . . . . . . . . . . . . . . . . . 121.8 Inžuence of environment on viral communities . . . . . . . . . . . . . . 151.9 Viral and microbial dynamics during ecological disturbances . . . . . . 161.10 Choice of genetic targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171.11 Caveats (methodological considerations) . . . . . . . . . . . . . . . . . . 221.12 Research objectives and outline of thesis . . . . . . . . . . . . . . . . . . . 251.13 Signicance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 High temporal and spatial diversity in marineRNA viruses implies thatthey have an important role in mortality and structuring planktoncommunities. . . . . . . . . . . . . . . . . . . . . . . . . . . . 292.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38v2.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533 Marinevirusandhostcommunity structure exhibits temporal phylogeneticdynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 553.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 583.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 693.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 934 Network analysis of Jericho Pier microbial time-series . . . . . . . . 994.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 994.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1004.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1044.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1114.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1265 Viral and heterotrophic protistan control of a phytoplankton bloom . 1335.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1335.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1345.3 Material and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1385.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1455.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1585.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1626 Conclusion and Future Directions . . . . . . . . . . . . . . . . . 1646.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1646.2 Additions to the “seed bank” theory . . . . . . . . . . . . . . . . . . . . . 1666.3 e nature of ephemeral and persistent OTUs over time . . . . . . . . . 1676.4 Ešect of environmental parameters on determining the co-occurrenceof viruses and hosts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1676.5 Ešect of disturbances on microbial communities . . . . . . . . . . . . . . 1686.6 Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1696.7 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1706.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . 173Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202a Supplementary Information to Chapter 2. . . . . . . . . . . . . . . 202a.1 Supplementary data and gures . . . . . . . . . . . . . . . . . . . . . . . . 202vib Supplementary Information to Chapter 3 . . . . . . . . . . . . . . . 207b.1 Supplementary gures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207c Supplementary Information to Chapter 4. . . . . . . . . . . . . . . 229c.1 Supplementary gures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229viiList of TablesTable 2.1 Description of samples and resulting sequencing information. . . . 39Table 2.2 Phylogenetic diversity, species richness. . . . . . . . . . . . . 45Table 3.1 PCR parameters used in this study. . . . . . . . . . . . . . . 60Table 3.3 Spearman correlations among environmental parameters, communityrichness, and community similarity. . . . . . . . . . . . . . 83Table 3.9 Mantel tests among community similarity matrices and distancematrices of environmental data. . . . . . . . . . . . . . . . 88Table 4.1 Network statistics. . . . . . . . . . . . . . . . . . . . . 120viiiList of FiguresFigure 1.1 Marine food web adapted from Azam et al. (1983), Wilhelm andSuttle (1999) and Worden et al. (2015). . . . . . . . . . . . . . 3Figure 2.1 Location of sampling sites. . . . . . . . . . . . . . . . . . 33Figure 2.2 Environmental parameters. . . . . . . . . . . . . . . . . . 40Figure 2.3 Rarefaction curves. . . . . . . . . . . . . . . . . . . . . 42Figure 2.4 Euler diagrams of normalized RdRp OTUs.. . . . . . . . . . . 43Figure 2.5 Rank abundance by site. . . . . . . . . . . . . . . . . . . 44Figure 2.6 Phylogenetic tree with heatmap. . . . . . . . . . . . . . . . 46Figure 3.1 Chlorophyll a concentration over time at Jericho Pier, Vancouver,British Columbia. . . . . . . . . . . . . . . . . . . . . . 67Figure 3.2 Environmental parameters during 1-year time series at Jericho Pier. . 68Figure 3.3 Rarefaction curves of samples from Jericho Pier time series. . . . . 71Figure 3.4 Species richness, phylogenetic diversity, and community similarityover time. . . . . . . . . . . . . . . . . . . . . . . . . 72Figure 3.5 Maximum likelihood RAxML phylogenetic trees and barplots ofclosely-related phylogenetic groups of OTUs. . . . . . . . . . . 73Figure 3.6 Maximum likelihoodphylogenetic tree (RAxML) ofmarine picorna-like viruses including reference sequences and OTUs. Outgroup isvirus Equine rhinitis B virus (Picornaviridae). . . . . . . . . . . 74Figure 3.7 Maximum likelihoodphylogenetic tree (RAxML) ofT4-likemyovirusesincluding reference sequences and OTUs. . . . . . . . . . . . 75Figure 3.8 Barplot of top 20most relatively abundantmarine picorna-like virusOTUs. . . . . . . . . . . . . . . . . . . . . . . . . . 76Figure 3.9 Barplot of the top 20 most relatively abundant T4-like myovirusOTUs. . . . . . . . . . . . . . . . . . . . . . . . . . 77Figure 3.10 Bacterial OTUs by phylum over time. . . . . . . . . . . . . . 79Figure 3.11 Eukaryotic OTUs by phylum over time. . . . . . . . . . . . . 80Figure 3.12 Plot of relative abundance of T4-like myoviral OTUs (95% aminoacid similarity) thatwere present in over 90%of samples (persistent)or in less than 20% of samples (ephemeral).. . . . . . . . . . . 81Figure 3.13 Plot of relative abundance ofmarine picorna-likeOTUs (95% aminoacid similarity) thatwere present in over 90%of samples (persistent)or in less than 20% of samples (ephemeral).. . . . . . . . . . . 82ixFigure 3.14 Marine picorna-like viral group A compared to eukaryotic OTUsclassied as raphidophytes over time. . . . . . . . . . . . . . 90Figure 3.15 Marine picorna-like group A sequences aligned. Dišerences to themost abundant OTU are highlighted. . . . . . . . . . . . . . 91Figure 3.16 T4-like myoviral group I compared to bacterial OTUs classied ascyanobacteria over time. . . . . . . . . . . . . . . . . . . 92Figure 4.1 Legend for networks. . . . . . . . . . . . . . . . . . . . 112Figure 4.2 Bacterial and eukaryotic networks. . . . . . . . . . . . . . . 113Figure 4.3 Viral community networks. . . . . . . . . . . . . . . . . . 115Figure 4.4 Networks between dišerent communities. . . . . . . . . . . . 116Figure 4.5 Interactions over time, between and within communities. . . . . . 118Figure 4.6 Counts of triplets by environmental factor. . . . . . . . . . . . 119Figure 4.7 Degree histograms. . . . . . . . . . . . . . . . . . . . . 123Figure 4.8 Network statistics over time of the subnetworks composed of twodišerent communities andof subnetworkswithin single communities.124Figure 4.9 Variation partitioning based on partial RDA. . . . . . . . . . . 125Figure 5.1 Environmental parameters during summer algal bloom 21 June -5July 2011. . . . . . . . . . . . . . . . . . . . . . . . . 146Figure 5.2 Environmental parameters during 1 year time series at Jericho Pier. . 147Figure 5.3 Richness of amplicons during the summer algal bloom and overall13 month time series. . . . . . . . . . . . . . . . . . . . 148Figure 5.4 Evenness of amplicons during the summer algal bloom and overall13 month time series. . . . . . . . . . . . . . . . . . . . 149Figure 5.5 Top 20most relatively abundantOTUs fromeach community duringthe summer algal bloom and over the entire time series.. . . . . . 151Figure 5.6 Relative abundance of bacterial and eukaryotic OTUs classied byphyla, during the summer algal bloom (leŸ panels) and annually(right panels). . . . . . . . . . . . . . . . . . . . . . . 152Figure 5.7 Relative abundance of bacterial and eukaryotic OTUs classied byclass, during the summer algal bloom and annually. . . . . . . . 153Figure 5.8 Change in relative abundance of OTUs for four eukaryotic ordersduring the summer algal bloom and throughout the year. . . . . . 155Figure 5.9 Oligotypes of specic OTUs pf marine picorna-like viruses duringthe summer algal bloom and annually. . . . . . . . . . . . . 156Figure 5.10 Oligotypes of OTUs during the bloom and annually. . . . . . . . 157Figure a.1 Percent similarity vs. number of OTUs. . . . . . . . . . . . . 204Figure a.2 Control sequence reads clustered at 95% similarity. . . . . . . . . 205xFigure a.3 Control sequence and PCR amplied reads denoised at dišerentpercentages using the QIIME denoiser Titanium settings (Reederand Knight, 2010). . . . . . . . . . . . . . . . . . . . . 206Figure b.1 Legend for tip colours for Maximum likelihood trees for marinepicorna-like viruses. . . . . . . . . . . . . . . . . . . . . 207Figure b.2 Maximum likelihood phylogenetic tree (RAxML) of subsection Aof RdRp including reference sequences and OTUs generated in thisstudy. . . . . . . . . . . . . . . . . . . . . . . . . . 208Figure b.3 Maximum likelihood phylogenetic tree (RAxML) of subsections Band C of RdRp including reference sequences and OTUs generatedin this study. . . . . . . . . . . . . . . . . . . . . . . . 209Figure b.4 Maximum likelihood phylogenetic tree (RAxML) of subsections Dand E of RdRp including reference sequences and OTUs generatedin this study. . . . . . . . . . . . . . . . . . . . . . . . 210Figure b.5 Maximum likelihood phylogenetic tree (RAxML) of subsection Fand G of RdRp including reference sequences and OTUs generatedin this study. . . . . . . . . . . . . . . . . . . . . . . . 211Figure b.6 Maximum likelihood phylogenetic tree (RAxML) of subsection Hof RdRp including reference sequences and OTUs generated in thisstudy. Subsection views are for the Group H (in brown). . . . . . 212Figure b.7 Legend for tip colours for the Maximum likelihood tree of T4-likemyovirises. . . . . . . . . . . . . . . . . . . . . . . . 213Figure b.8 Maximum likelihoodphylogenetic tree (RAxML) of subsectionAofgp23 (marker forT4-likemyoviruses) including reference sequencesand OTUs generated in this study. . . . . . . . . . . . . . . 214Figure b.9 Maximum likelihood phylogenetic tree (RAxML) of subsection B ofgp23 (marker forT4-likemyoviruses) including reference sequencesand OTUs generated in this study. . . . . . . . . . . . . . . 215Figure b.10 Maximum likelihoodphylogenetic tree (RAxML) of subsectionCofgp23 (marker forT4-likemyoviruses) including reference sequencesand OTUs generated in this study. . . . . . . . . . . . . . . 216Figure b.11 Maximum likelihood phylogenetic tree (RAxML) of subsection Dand E of gp23 (marker for T4-like myoviruses) including referencesequences and OTUs generated in this study. . . . . . . . . . . 217Figure b.12 Maximum likelihood phylogenetic tree (RAxML) of subsection F ofgp23 (marker forT4-likemyoviruses) including reference sequencesand OTUs generated in this study. . . . . . . . . . . . . . . 218Figure b.13 Maximum likelihood phylogenetic tree (RAxML) of subsection G(top) of gp23 (marker for T4-like myoviruses) including referencesequences and OTUs generated in this study. . . . . . . . . . . 219xiFigure b.14 Maximum likelihood phylogenetic tree (RAxML) of subsection G(bottom) of gp23 (marker forT4-likemyoviruses) including referencesequences and OTUs generated in this study. . . . . . . . . . . 220Figure b.15 Maximum likelihood phylogenetic tree (RAxML) of subsection Hof gp23 (marker forT4-likemyoviruses) including reference sequencesand OTUs generated in this study. . . . . . . . . . . . . . . 221Figure b.16 Maximum likelihood phylogenetic tree (RAxML) of subsection I(top) of gp23 (marker for T4-like myoviruses) including referencesequences and OTUs generated in this study. . . . . . . . . . . 222Figure b.17 Maximum likelihood phylogenetic tree (RAxML) of subsection I(bottom) of gp23 (marker forT4-likemyoviruses) including referencesequences and OTUs generated in this study. . . . . . . . . . . 223Figure b.18 Heatmap of relative abundance of eukaryoticOTUs (97% similarity)over time. . . . . . . . . . . . . . . . . . . . . . . . . 224Figure b.19 Heatmap of relative abundance of bacterial OTUs (97% similarity)over time. . . . . . . . . . . . . . . . . . . . . . . . . 225Figure b.20 Heatmapof relative abundance of T4-likemyoviralOTUs (95% similarityamino acid) over time ordered by phylogenetic tree tree (Figureb.3.7). . . . . . . . . . . . . . . . . . . . . . . . . . 226Figure b.21 Heatmap of relative abundance of marine picorna-like OTUs (95%similarity amino acid) over time ordered by phylogenetic tree tree(Figure b.3.6). . . . . . . . . . . . . . . . . . . . . . . 227Figure b.22 Non-metric dimensional scaling (NMDS) plot of themicrobial communitiescoloured by season. . . . . . . . . . . . . . . . . . . . . 228Figure c.1 Overall network (includesOTUs from eukaryotic, bacterial, T4-likemyoviruses and environmental parameters) compared to simulatednetworks created from same number of nodes and edges. . . . . . 230Figure c.2 Network of eukaryoticOTUs compared to simulatednetworks createdfrom same number of nodes and edges. . . . . . . . . . . . . 231Figure c.3 Network of bacterialOTUs compared to simulatednetworks createdfrom same number of nodes and edges. . . . . . . . . . . . . 232Figure c.4 Network of marine picorna-like viral OTUs compared to simulatednetworks created from same number of nodes and edges. . . . . . 233Figure c.5 Network of bacterial andT4-likemyoviralOTUs compared to simulatednetworks created from same number of nodes and edges. . . . . . 234Figure c.6 Network of T4-likemyoviralOTUs compared to simulatednetworkscreated from same number of nodes and edges. . . . . . . . . . 235Figure c.7 Network of eukaryotic and bacterial OTUs compared to simulatednetworks created from same number of nodes and edges. . . . . . 236xiiFigure c.8 Network of eukaryotic andmarine picorna-like viralOTUs comparedto simulated networks created from samenumber of nodes and edges.237xiiiList of AcronymsAdonis Analysis of variance using distance matrices.ARISA Automated Ribosomal Intergenic Spacer Analysis.AVS Algal virus specic, Glossary: AVS.BC British Colombia.BLAST Basic Local Alignment Search Tool.chl a Chlorophyll a.CRT Conditionally rare taxa.DGGE Denaturing gradient gel electrophoresis.e-value Expect value.gp23 Gene product 23, Glossary: gp23.JP Jericho Pier.LSA Local similarity analysis.NGS Next Generation Sequencing, Glossary: NGS.NMDS Non-metric multidimensional scaling, Glossary: NMDS.OTU Operational taxonomic unit.xivPCR Polymerase chain reaction.PFGE Pulse eld gel electrophoresis.qiime Quantitative Insights Into Microbial Ecology.RAxML Randomized Axelerated Maximum Likelihood.RdRp RNA dependent RNA polymerase, Glossary: RdRp.SOG Strait of Georgia.T-RFLP Terminal Restriction Fragment Length Polymorphism.xvGlossaryα, β diversity e species diversity (or richness) of a local community or habitat (α di-versity); the dišerence in diversity associated with dišerences in habitat or spatialscale (β diversity).abiotic Having to do with the chemical, geological, and physical aspects of an entity;i.e., the nonliving components.amplicon sequencing Sequencing of all the variants in a targeted region of a genomeamplied using PCR. In this dissertation the primers used to amplify the targetedregion were designed to encompass a large proportion of the communities.usthe sequence reads are frommany dišerent species/strains amplied from the sam-ple. Also known as: amplicon deep sequencing, ultra-deep sequencing, eDNA,pyrotags.AVS Algal virus specic primer set. Described in Chen and Suttle (1995).Bacillariornavirus Viral genus of ssRNA viruses infecting diatoms. In the viral orderPicornavirales.bank theory Within a community of viruses or cells there are a few abundant viruses,however, most viruses are rare. ese rare viruses can rapidly become abundantbased on the environment (or hosts or other interactions) that are available. Alsoknown as: seed bank theory.biogeochemistry e scientic study of the physical, chemical, geological, and biolog-ical processes and reactions that govern the cycles of matter and energy in thenatural environment.xvibiotic Having to do with or involving living organisms.bloom Apopulation outbreak ofmicroscopic algae (phytoplankton) that remainswithina dened part of the water column.bottom-up control e regulation of ecosystem structure and function by factors suchas nutrient supply and primary production at the base of the food chain, as op-posed to “top-down” control by consumers.burst size Number of viruses produced per infected cell.capsid Protein shell of viruses composed of many subunits.Caudovirales Viral order of double-stranded DNA viruses that includes the tailed bac-teriophage.cloning Replication of DNA inside bacterial cell using a vector into which the targetDNA has been placed.coevolution Aprocess of reciprocal evolutionary change in two interacting species, drivenby natural selection.coexistence e indenite persistence of two or more species within the same commu-nity; this involves species that will continue to persist in the face of perturbationsin their abundances. Species that co-occur may or may not be stably coexisting,because one or more of themmay be on the way to local extinction at a time scalethat is too slow to be immediately apparent.community An assemblage of species found together in a specic habitat at a certaintime.community structure Determined by species composition and relative abundance.xviidegree In graph theory, the degree of a node is how many edges are connecting thisnode to other nodes.direct ešect e immediate impact of one species on another’s chance of survival andreproduction, through a physical interaction such as predation or interference.disturbance An episodic event that results in a sustained disruption of an ecosystem’sstructure and function. is may be a physical disturbance, a biological distur-bance, or an anthropogenic disturbance.dynamics e changes through time in the size of a population, or in a related measuresuch as density.edge In graph theory, refers to the connection between two nodes in a network. estrength of the connection can be displayed. Also called: link.endemic A species that has a relatively narrow geographic range, such as one that isfound only in a particular body of water or in a particular habitat or region.ephemeral Short time period, also: transient.evenness Similarity in number or in proportion of the species in a community.ngerprinting Molecular biology techniques used to quickly survey the genetic diver-sity of a sample.gp23 Gene product 23. Codes for the major capsid protein in some families of bacterio-phage.heterotroph An organism that must consume organic compounds as food for growth.homologues Genes with shared ancestry.xviiiIllumina Company that makes High-throughput sequencers such as the Miseq, Hiseqand NextSeq.Killing the Winner Frequency-dependent selection onhost types asmediated by viruses."Winners" are hosts that have achieved su›cient numbers that they are capableof supporting virus population growth to densities that can drastically reduce innumber these same virus-susceptible hosts.Labyrnavirus Viral genus of ssRNA viruses infecting thraustrochytrids. In the viralorder Picornavirales.library Collection of genetic material prepared for sequencing.Lotka Volterra Model composed predator–prey dynamics. Model composed of twoequations that describe the dynamics of biological systems in which two speciesinteract, one as a predator and the other as prey.e model describes oscillationsin the population size of both predator and prey, with the peak of the predator’soscillation lagging slightly behind the peak of the prey’s oscillation.Marnaviridae Viral family of ssRNA viruses infecting Raphidophyte alga. In the viralorder Picornavirales.mesocosm Experimental water enclosures. outdoor experiment of the natural environ-ment that is controlled. Examples: 60L bags can be used for aquatic experiments,can be žoated in-situ, but it is a closed system.microbe Another term for a microorganism.microeukaryotes Microscopic eukaryote.mineralization e microbially mediated conversion of organically bound nutrientssuch as nitrogen and phosphorus to soluble inorganic forms that can be taken upby plants.xixMiseq A type of Illumina sequencing platform.monophyletic Describing a group of species that are more closely related to each otherthan any of them are to other species outside the group.us, monophyly.Myoviridae Viruses infecting bacteria and archaea that are members of the viral orderCaudovirales.NGS Next Generation Sequencing. High-throughput sequencing. Generally refers todata from Illumina and Roche 454 sequencers.niche e specic role and requirements of a particular population or species within alarger community.NMDS Non-metric multidimensional scaling. An ordination used to visualize the vari-ation from multiple variables in 2-3 dimensions.node In a network, refers to an object that has been compared to the other objects. Inthis dissertation a node is an OTU or an environmental parameter. Also called:vertex.partial redundancy analysis Analysis like redundancy analysis, but can try removingthe ešect of one variable on the summarised variation.persistent Enduring for a long period.Picornavirales Viral order of positive-sense ssRNA viruses.plankton A collective term for various driŸing organisms of the pelagic zone. Phyto-plankton are photosynthetic primary producers, and zooplankton are consumers.Podoviridae Viral family of Caudovirales with short tailed bacteriophage.population A group of individuals of the same species occupying a certain geographicarea over a specied period of time.xxproductivity Rate of generation of biomass in an ecosystem.prokaryote Asingle-celled organism lackingmembrane-bound anucleus, and organelles.Not the preferred term because it implies that there are only two types of organ-isms.protist Unicellular eukaryote.protozoa Unicellular eukaryotic non-photosynthesis organism.pyrosequencing High-throughput sequencing method following "sequencing by syn-thesis" model. Complementary strands are formed and each time a new base isadded light from the reaction is detected.e most common type of this sequenc-ing was Roche’s 454 sequencing.quasispecies Group of viruses related by mutations. Viral "species" is oŸen an averageof all of these mutant (relative to ancestral) sequences.Q value Minimum false discovery rate where the test is deemed signicant.rank abundance curve Plot type used to display relative species abundance. X axis isthe rank of the species and Y-axis is the relative abundance of the species.rare biosphere Where most species are rare and few are abundant. Recent discussionhas been precipitated by high-throughput sequencing approaches.rarefaction curve e statistical expectation of the number of species in a survey orcollection as a function of the accumulated number of individuals or samples,based on resampling from an observed sample set.RdRp RNA dependent RNA polymerase. Enzyme that catalyzes the replication of RNAfrom RNA.is enzyme is generally well-conserved in viruses, but also found ineukaryotes as part of some RNA Interference pathways. It has not been found inbacteria and archaea.xxiread normalizing e number of reads from dišerent libraries must be made similar.is is done to account for uneven reads per library as a result modern sequencingtechniques. Each library is rareed (reads picked randomly without replacement)to the number of reads from sequencing library with the lowest number of reads.reads Sequence reads from DNA sequence, usually refers to raw or quality trimmedsequences.redundancy analysis Technique to summarise the variation in object (in this disserta-tion OTUs or groups of OTUs) by another set of explanatory variables.relative abundance e quantitative pattern of rarity and commonness among speciesin a sample or a community.resilience e ability of an ecosystem to recover from or resist disturbances and pertur-bation, so that the key components and processes of the system remain the same.Sanger sequencing Original sequencing platformwhere sequence is determined by chainterminating dideoxynucleotides that are incorporated during a replication withDNA polymerase. Low throughput, but low errors and long read length.Siphoviridae Viral family of Caudovirales with long tailed bacteriophage.species area curve Relationship between the number of species found and the area. alsoknown as: ešort curve.species richness e number of species in a community, or in a region.top-down control Regulation of ecosystem structure and function by consumers ratherthan factors such as nutrient supply and primary production at the base of the foodchain.trophic level e position of a given species in the chain of energy or nutrients.xxiiAcknowledgementsIt truly "takes a village" to complete a PhD and while I cannot thank everyone who hashelpedme, I would like to take a bit of space to try to thank a few of the people who havehelped me along the way.I would like to acknowledge the support and trust of my supervisor, Curtis Suttle,throughout this journey.My supervisory committee, Steven Hallam, François Jean and Maite Maldonado,have challenged my thoughts, experiments and analysis during my PhD. I am gratefulfor their high expectations and their help.I would like to thank all Suttle lab members and visitors past and present for theirhelp, constructive criticism, and laughs. I would especially like to thank Amy Chan,Renat Adelshin, Anwar Al-Qattan, Christina Charlesworth, Caroline Chénard, AnnaCho, Cheryl Chow, Jessie Clasen, Jenn Cook, Chris Deeg, Jan Finke, Matthias Fisher,Jingze Jiang, Jessica Labonté, John Li, Jérôme Payet, Emma Shelford, Alvin Tian, RickWhite III, Danielle Winget, Christian Winter, Jennifer Wirth, Marie-Claire Veuilleux,Marli Vlok, and Kevin Zhong. anks to Chris Payne for processing nutrient samples,giving guidance on cruise-prep and general ocean-going expertise, and thanks to RichPawlowicz for use of his lab’s YSI.Many people have helped me with editing, writing and presentations. Some stand-outs include: Caroline Chénard, Cheryl Chow, Jessie Clasen, Chris Deeg, Jan Finke,Brenda Gustavsen, ierry Heger, Elisabeth Hehenberger, Bernhard Konrad, EmmaShelford, Xi (Alvin) Tian, Marli Vlok, Xu (Kevin) Zhong.I would also like to thank SoŸwareCarpentry for providing the best timed-workshopof my doctoral degree. anks to Greg Wilson, for being a good mentor, and to JennyBryan for enriching opportunities.xxiiiI have received generous funding while at UBC which made it possible to completemy studies. I have been supported by a National Sciences and Engineering ResearchCouncil PGS-M, a PGS-D, a UBC Four Year Fellowship, a UBC Earth, Ocean and At-mospheric Sciences scholarship top-up, and a Beaty Biodiversity Biodiversity ResearchIntegrative Training and Education travel award. e lab has been supported by Cana-dian Institute for Advanced Research, NSERC discovery grant and CFI.anks to all the friends who had time for cošee, beer, dinners, hikes, bike touring,camping, backpacking trips, skiing, swimming, and yoga.I would also like to thank my parents and stepmom for their continued support andencouragement as I pursued this degree four time zones and 4000 kilometers away fromthem. Finally, I thankierry Heger for his overall support, ideas, editing, much loveand kindness, and zest for adventures that helped me to keep going throughout thisdegree.xxivchapter 1Introduction1.1 synopsis1.1.1 Overall importance of marine virusesViruses are obligate, intracellular parasites. ey range in diameter from 20 nm to 750nm (Fuhrman, 1999; Arslan et al., 2011), and have genome sizes from 2 kb (Circoviri-dae) (Gorbalenya et al., 2006) to 2.5 Mbp (“Pandoravirus”) (Philippe et al., 2013). eycontain either DNA or RNA as genetic material which is used to hijack host replicationfor the production of new viruses. In the ocean, viruses have high abundances (Berghet al., 1989; Proctor and Fuhrman, 1990; Suttle et al., 1990), a cosmopolitan distribution(Angly et al., 2006; Liang et al., 2014),and play an important role through their ešecton biogeochemical cycles (Fuhrman, 1999; Wilhelm and Suttle, 1999). Aquatic virusesharbour some of the highest genetic diversity on earth (Suttle, 2007; Brum and Sullivan,2015). Molecular surveys of viral marker genes (Short and Suttle, 2002; Payet and Suttle,2014) and metagenomic surveys of viral genomes (Breitbart et al., 2002; Culley et al.,2006; Winget and Wommack, 2008) have enabled glimpses of this high diversity anddynamics under changing conditions.1.1.2 Temporal dynamics of aquatic viruses and microbesViruses žuctuate over time in abundance and community composition (Hewson et al.,2006b; Needham et al., 2013). Viruses show annually and seasonally repeatable patternsin abundance (Parsons et al., 2012). Compositionally, marine bacteriophage commu-nities are more similar across seasons than years (Chow and Fuhrman, 2012; Marston1chapter 1and Sallee, 2013) and are more similar among connected waters than among isolatedbodies of water (Marston and Sallee, 2013). Temporal dynamics have been examinedextensively in bacterial communities using the small subunit (SSU) of the ribosomalRNA (rRNA) gene 16S (Horner-Devine and Bohannan, 2006; reviewed in Nemergutet al., 2013) and to a lesser extent in eukaryotic communities using the 18S rRNA gene(Massana et al., 2015). Despite these studies, very little is known about the dynamicsof viral community composition and structure across multiple viral families and thephylogenetic-relatedness of viral communities during stochastic or repeatable events inthe environment.To gain more insights into viral community structure and dynamics, I surveyedchanges in viral and putative host communities using a year-long study at a coastalsite. Changes over timewere examined every twoweeks using high-throughput sequenc-ing of well-established marker genes of ecologically important viral families, and withdomain-specic primers of bacteria and eukaryotes. ese datasets will illuminate theimportant role of viruses over time as modulators of the environment. More specically,the viral phylogenies will elucidate the identity and ecology of these viruses, and howviral phylogeny could be related to host phylogeny. e community structure demon-strates how communities of viruses and other organisms are regulated.Marker genes come with certain caveats, and this will be discussed later in section1.11, but targeting specic communities, is invaluable as the community can be assessedmore deeply than with metagenomics (random sequencing of all genetic viral material).Initial feasibility of this approach was done using two timepoints and three spatiallyproximate samples (Chapter 2). e overall approach of the dissertation is unique inthat it combines a holistic view of microbial and viral communities, plus it uses high-throughput sequencing to gain insights about these communities.2chapter 11.2 microbial influence on overall food web1.2.1 What is the microbial loop?Microbes are responsible for recycling large amounts of the material in aquatic ecosys-tem (Azam et al., 1983; Worden et al., 2015) (Figure 1.1). Microbes cycle the dissolvedorganicmaterial (DOM)up to higher trophic levels by taking upDOM.Whenorganismsdie, material is released and the DOM is recycled when it is taken up by microbes.Figure 1.1: Marine food web from Azam et al. (1983), Wilhelm and Suttle (1999) andWorden et al. (2015).1.2.2 Viral shunte classical food web, which ignored microorganisms and viruses, was changed withthe addition of viruses because the carbon and energy that would be transferred tohigher trophic levels is “shunted” to lower trophic levels as viruses lyse organisms whichwill subsequently release their cellular contents (Wilhelm and Suttle, 1999; Weitz andWilhelm, 2012).ismaterial can then be taken up by themicrobial loop and transferredup the food web.3chapter 11.3 why study temporal dynamics?Determining how organisms change over time has long been an important ecologicalquestion. e consideration of time has shaped many ecological theories such as Dar-win’s ideas about the origin of species and processes involved in Island Biogeography(MacArthur and Wilson, 1967). Community dynamics help explain more about theecology of organisms and perhaps even what drives their diversity. Furthermore, somedynamics are only visible over long timescales. For example, bacterial communities canbe predictable at a monthly scale, but not at a daily scale (Fuhrman et al., 2015). Muchcan be learned about the ecology of organisms by examining their dynamics over time,and how they change together or with the environment or both (Levin, 1992; Chessonand Huntly, 1997).1.3.1 Models describing viral and microbial dynamicsAlthough it is intuitive that viruses inžuence host populations through cell lysis, mostmarine viruses’ hosts are unknown. Viral inžuence on the host communities can be verysubstantial, as viruses play a large role in killing hosts. As well, they exert selective forceson the host community composition. Viruses may increase the diversity of bacterialcommunities (Bouvier and del Giorgio, 2007;Middelboe et al., 2009). Another potentialmechanism is reduction in number ormutations in bacterial receptors, thusminimizingvirus-receptor contact rates (Lenski, 1988).Many predator-prey or virus-host relationships have been described using Lokta-Volterra models (Holt and Pickering, 1985). ese models describe how the abundanceof a predator lags behind the rise in abundance of the “prey” or host.e Lokta-Volterramodel inspired many theories governing dynamics between hosts and parasites, includ-ing one popular in viral ecology called “Killing the Winner” (ingstad, 2000; Winteret al., 2010;ingstad et al., 2014). e “Killing the Winner” hypothesis states that forprokaryotes, in addition to constant protozoan grazing, viruses will kill themost quicklygrowing prokaryote (via increased encounter rate) therefore promoting host community4chapter 1succession and diversity (ingstad, 2000; Winter et al., 2010; Storesund et al., 2015).us, dišerent viruses predominate at dišerent times and the host community turnoverpromotes diversity within viruses. However, strong evidence from laboratory experi-ments and eld studies is sparse since some studies show these dynamics (e.g. Hewsonet al., 2003; Schwalbach et al., 2004; Rodriguez-Brito et al., 2010) while others do not (e.g.Hewson et al., 2006b).is theory can also be adapted to examine how the strain-levelspecicity is operating to maintain the diversity of the system (ingstad et al., 2015).Another development from the Lokta-Volterra theories includes the “Red Queen”hypothesis. is hypothesis states that a host is always doing as much as it can to es-cape the predatory pressure of a parasite and to remain in its current ecological space(VanValen, 1973; Benton, 2009). With the discovery of the viruses infecting SAR11 (Zhaoet al., 2013), a bacterial clade containing some of the most dominant marine bacteria,there have been new theories proposed to explain SAR11’s continued persistence in spiteof the presence of viruses. One such idea is the “King of the Mountain” theory namedin Giovannoni et al. (2013)’s reply to Våge et al. (2013).is theory examines dišerencesamong hosts and the “King of theMountain” (KoM) is the superior resource competitorin a system. us, with high abundances, the population avoids being decimated byviruses. is forms a positive feedback loop. Consequently, these “Kings” contributemore to geochemical cycles than the lesser competitors.1.4 diversity over time1.4.1 Measurements of richnessere are several ways to describe how viruses and microbes are distributed across en-vironments. For example, the count of dišerent species or operational taxonomic units(OTUs -usedwithmicrobes to approximate a division of an organism based on sequenceidentity) in a sample is the richness, which can be used to compare communities fromdišerent environments and sampling dates. Moreover, diversity is calculated from both5chapter 1the richness, and the relative abundance of species (evenness) in a sample usingmethodssuch as the Shannon diversity index (Shannon, 1948), and the Simpson index (Simpson,1949). For most organisms the richness and diversity of communities may be inžuencedby factors such as latitude, productivity, and habitat (Gaston, 2000). For viruses, hostslikely inžuence their distribution and genetic diversity (Mizumoto et al., 2006; Yang et al.,2010), yet, with more than one type of virus infecting one host, and some viruses infect-ing multiple types of hosts (e.g. Sullivan et al., 2003), the hosts might not be singularlydetermining viral diversity.1.4.2 How is richness and diversity measured in viral and microbial communities?Oneway to examine the richness of viral andmicrobial communities is by usingmolecu-lar techniques. Before the advent of high-throughput sequencing (HTS) of PCR-ampliedmarker genes, researchers usedmolecular ngerprintingmethods such as denaturing gelgradient electrophoresis (DGGE)(e.g. Frederickson et al., 2003; Payet and Suttle, 2014),and terminal restriction fragment length polymorphism (T-RFLP) (e.g. Pagarete et al.,2013; Chow et al., 2014) to quantify the diversity of viral genotypes. To examine the diver-sity of the overall communities based on viral genomes, pulsed-eld gel electrophoresis(PFGE) (e.g. Steward et al., 2000) was used. PFGE is very low resolution because therecould be many overlapping sizes of viral genomes.ese methods provided very usefulcomparisons of samples, but were limited to themost abundant members of the commu-nities. Furthermore, the sequences of the community members were unknown unlessadditional steps were performed such as cutting out bands from the gel, cloning, andSanger sequencing.ese techniques have been used similarly in bacterial (Crump et al.,2004) and eukaryotic communities (Diez et al., 2001). Much of this work has moved tousing high-throughput sequencing (a.k.a. amplicon sequencing) of the PCR-ampliedmarker genes.6chapter 11.4.3 Temporal dynamics of viruses and putative hosts at Jericho PierLooking for temporal shiŸs in viral communities, Short and Suttle (2003) examined acoastal site, Jericho Pier (JP), BC, at one-week intervals for 14 months. Using markergenes for algal viruses (AVS) and potential hosts (18S rDNA), they examined changesin the community composition using DGGE.e viral community was relatively sta-ble throughout time, whereas the potential host community showed greater temporalvariation. is demonstrates that environmental changes have a greater ešect on mi-croeukaryotes than on viruses, or that the viral primers were specic for a small groupof viruses, and the potential host primers amplied a less specic group of eukaryotesand thus showed greater žuctuations.e shiŸs in viral composition were uncorrelatedto the hosts, however, some shiŸswere associatedwith tide height, salinity or chlorophylla concentration.1.5 community structure of microbial communitiesAmplicon sequencing andmetagenomic ešorts directed towards marine bacteria (usingthe 16S rRNA gene) have revealed the possibility of more bacterial species than previ-ously recognized (Sogin et al., 2006; Brown et al., 2015). Amplicon sequencing revealedthese communities were dominated by a small subset of genotypes, but most genotypeswere rare (Pedrós-Alió, 2012). Furthermore, in a controlled freshwater aquaculture lakeand in three solar salterns in Southern California, the abundant microbes and DNAviruses persisted over time, however, the rarer members of the community were notdetectable in every sample based on the metagenomic surveys (Rodriguez-Brito et al.,2010). e žuctuations in both microbial and viral communities showed a shuœing ofdominant types, but generally no extinguishing of specic viral genotypes. is sug-gests that although viral communities tend to be dominated by few genotypes, they stillmaintain high richness. ese examples of dišerent spatial and temporal žuctuationsin the composition of viral communities lead to the question of whether, for viruses,7chapter 1“everything is everywhere, the environment selects,” or if there are endemic viruses.erefore, examining viral communities over time, and during dišerent environmentalconditions will illuminate how these factors inžuence viral diversity and communitycomposition.1.5.1 Uneven structure of viral communitiese structure of viral communities, i.e. how their genotypes are distributed in an envi-ronment, appears to be very uneven (for bothDNAandRNAviral communities) (Culleyet al., 2006; Djikeng et al., 2009; Rodriguez-Brito et al., 2010). In RNAviral communitiesat twoBritishColumbian sites, Jericho Pier (JP), and in the Strait ofGeorgia (SOG),mostsequences had no homologues and were rare (Culley et al., 2006).e two communitieswere dominated by dišerent genotypes and therewas no sequence overlap between thesetwo communities, however, it must be noted that only 277 sequences were recovered,limiting the coverage of this study. Similarly uneven were RNA viral communities ina lake near Rockville, MD, sampled in June and November (Djikeng et al., 2009). elake was dominated by 11 genotypes and many more rare genotypes. Some viruses wereabundant at both sampling dates, however, some viruses were abundant one month andthen low in abundance or undetectable in the other month. ese changes in commu-nity composition were not examined in the context of potential hosts or environmentalparameters which could have correlated with community changes.1.5.2 Seed bank community structuree unevenness in viral communities is described by the Bank model (or seed bankmodel) of viral community structure (Breitbart and Rohwer, 2005).e Bank model isformed from the seed bank idea where the rank abundance curves from a communityfollow a log-normal type distribution (Chow and Suttle, 2015). Within the communitythere are a few abundant viruses, however, most viruses are rare. ese rare viruses8chapter 1form a “seed bank” where they can rapidly become abundant based on the environmentor hosts that are available.1.5.3 Temporal shiŸs in bacterial and eukaryotic community compositionMarine bacterial and eukaryotic communities are dynamic over time with shiŸs relatedto changes in day length (Gilbert et al., 2011) and composition related to salinity (Lozuponeand Knight, 2007). Viruses are dependent on their hosts and thus inžuenced by theavailability of hosts, however, as stated earlier (p. 4) these viruses also exert a selectivepressure on their host communities. Shade et al. (2014) found that conditionally raretaxa (CRTs), that is those that are usually rare but can quickly become abundant, can bedeemed responsible for many of the temporal dynamics seen in bacterial communities.It is has not been investigated how these dynamics relate to viral communities, however,it could be that viruses in the “Bank” lyse these CRTs following “Killing theWinner” typedynamics.1.5.4 OTU clusteringHigh-throughput sequencing methods generate large amounts of sequence data. To beable to analyse the data appropriately and to correct for sequencing error, sequencesare clustered into operational taxonomic units (OTUs), meaning they are grouped to-gether based on their similarity. ere are many dišerent types of sequence clusteringalgorithms (reviewed in Schmidt et al. (2015)), however, many have their roots in thehierarchical clustering algorithms used in community analysis such as single linkageand nearest neighbour clustering.e general process is that from all the sequence reads the unique sequences arechosen (speeds things up computationally), then ordered by length. Next, sequencesare compared, and if they are more than a set percentage similar they are grouped intoan OTU.en the next unique read is compared to this OTU and others, if the read issimilar (specically determined based on the algorithm) to an OTU already seen it is9chapter 1included in this OTU. If it is less similar than the cut-oš it will form a new OTU.eprocess continues until all of the unique reads are clustered into OTUs. For bacteria andeukaryotes the cut-oš of 97% sequence similarity is used for these studies to approxi-mate species level clusters (Stackebrandt and Goebel, 1994). e method used in thisdissertation is USEARCH (Edgar, 2010)– a centroid based approach for calculating thesimilarity to the sequence read.e OTUs are then used for further analyses requiringthe DNA sequence (e.g. phylogeny, taxonomic classication, etc.).1.5.5 Taxonomic classicationOTUs can be determined to resemble a reference sequence. is is called “classica-tion.” e Ribosomal database project (RdP) created a classier (Wang et al., 2007).e classier uses a naive (meaning independent) Bayesian approach by evaluating thefrequency of sequences cut up into a specic word size (8 was used in the paper becauseof memory constraints, but also for specicity).us each OTU and reference sequenceare represented by a collection of the counts of “words.”is allows the quick matchingof OTUs to the reference database and also gives condence limits to the assignment atdišerent levels of taxonomic specicity (Wang et al., 2007).OTUs are classied using a database of reference sequences that have a specic taxo-nomic assignment. For rRNA genes popular databases are Greengenes (DeSantis et al.,2006), RdP (Cole et al., 2014), Silva (Quast et al., 2013) and for 18S rRNA the ProtistRibosomal Reference database (PR2) (Guillou et al., 2013). For viral sequences there isno such database since there are comparatively few sequenced viruses, and do not haveshared genes among them.1.5.6 Phylogenetic context of OTUsTo compare genetic distances, phylogenetic methods can be used to examine OTUs.is has the advantage of not being reliant on a reference database, however, the out-puts are dišerent and can be complementary to the taxonomic classication when both10chapter 1are available. Taxonomic classication, however, is not generally based on phylogeny.Phylogeny-based approaches use the context of sequences on a phylogenetic tree as away to give a context to OTUs. Phylogenetic trees based on large amounts of sequencereads require the alignment of reads usingmultiple sequence alignments.e sequencesare either only used tomake a tree and then the genetic distances of OTUs are examined,or they can be added to the tree along with reference sequences. As well, they can beplaced onto tips of a well-characterized reference tree using a placement algorithm suchas evolutionary placement of short reads (EPA)(Berger et al., 2011) or pplacer (Matsenet al., 2010). ese placement algorithms are useful with well-dened reference treesand short sequences.OŸentimes, if all theOTUs are used (and there can be thousands), the trees are eithernot visualized or are heavily collapsed. A useful metric using phylogenetic distance isUnifrac (Lozupone and Knight, 2005) which compares the phylogenetic similarity ofcommunities, although this distancemetric is not without some issues (Long et al., 2014;Lozupone and Knight, 2015).1.6 relatedness over time1.6.1 Phylogenetic relatedness over timeBacterial communities aremore phylogenetically diverse than expected over time (Horner-Devine and Bohannan, 2006). Little is known, however, about the phylogenetic diversityof viral communities neither in general and over time, nor how it is maintained. Phylo-genetic diversity is usually related to species richness but little is known about why it canbe decoupled, only that it can happen under high levels of regional richness (Tucker andCadotte, 2013). Phylogenetic relatedness can be correlated to ecological relatedness inother organisms (Harvey and Purvis, 1991; Srivastava et al., 2012).11chapter 11.7 host-virus interactions over time1.7.1 Co-occurrence networksOrganisms can co-occur over time, and the more oŸen this happens, the more likelythere is a type of relationship (e.g. a symbiosis, or the same preferred niche). Whenspecies do not occur together theymay have an antagonistic relationship, either throughallelopathy, predation, or dišerent preferred niches. All of these co-occurrences can beexamined as a network. A network is a diagram of relationships, whereby the propertiesof all the pair-wise associations are examined together and the emergent properties ofthe overall interactions in the communities can be analysed.For microbial associations, three steps are required to generate networks. First, amatrix is generated by pairwise comparison of all organisms or those that are mostabundant. Many dišerent kinds of species abundance and molecular data can be com-pared with co-occurrence/correlation analysis such as presence-absence, species counts,relative abundance, high-throughput sequencing, DNA ngerprinting patterns, and mi-croarray data. ese pairwise association matrices can be generated using a variety ofdišerent methods including Spearman or Pearson correlation coe›cient, any distancemetric (e.g. Bray-Curtis, Euclidean, etc.), linear regressions (Faust and Raes, 2012), localsimilarity analysis (LSA) (Ruan et al., 2006; Xia et al., 2011) which can include time-lagged relationships or others. LSA can be used with time series data to look for thestrongest correlation among timepoints with a set maximum amount of lag.Second, the pairwise associations are ltered for strength of association or signif-icance or both. e signicance can be tested by examining the p-value (if available,oŸen calculated in pairwise associations by permutation, i.e. rearranging the matrixmany times and testing the results), and also by using the Q value which tests for falsediscoveries when the p-value is deemed signicant (Storey et al., 2005).ird, the signicant associations are assembled together into a network which canbe analysed for its emergent properties and also can be visualized. ese networks,also called graphs (a collection of pairwise relationships between objects), consist of12chapter 1two parts: nodes and edges. Nodes are the objects or variables used in the pairwisecomparison. When nodes are visualized they are drawn as points (can also be calledvertices). Edges are the associations between nodes and are oŸen drawn as lines (canalso be called links, connections, correlations). Edges can represent positive or negative,time-lagged, strong or weak associations. Properties of the individual nodes that can bedetermined are the degree of each node, which is number of nodes it is connected to byits edges.e properties of the overall network can be considered, which can be useful forcomparing networks (Cram et al., 2013). e overall degree distribution of the nodes,(e.g. how many have degree n) can be useful for classifying networks. For example,the betweenness, which is the number of shortest paths going through a node, can becalculated for one node, and also as an average over the whole network. e networkdensity (edges per node), the network diameter, and the clustering coe›cient (numberof groups of three nodes) can be examined. Triangles or triplets, which are three nodesand their edges can be used as a way to examine specic groups of interactions. Finally,modules, which are structures within the graph that are highly connected (oŸen biolog-ical networks show a high degree of modularity), can be detected.Networks can be compared to each other using some of the aforementioned proper-ties, although it is also oŸen useful to compare these networks to randomly generatednetworks (as a way to classify networks and to test for dišerence from randomly gener-ated network). Random networks can be generated based on some algorithms followingnetwork properties such as the scale-free (where degree distribution follows the powerlaw, wheremany vertices have a degree greater than average) and small-world. Generallymost microbial networks are classied as small-world networks where most nodes arenot neighbours of one another, but most nodes can be reached from every other nodethrough a small number of other nodes. Network visualization has been very popular inrecent years, however, care must be taken when interpreting biological meaning fromthe layout of a network. Networks can oŸen resemble “hairballs.” ey also can be13chapter 1visualized using force-directed layouts which attempt to lay out the nodes and edgesso that the edges are all about the same length and all nodes can be seen.At the SanPedroOceanTime series researchers used network analysis of viruses, bac-teria and protists to look for associations in theirmonthly coastal time series (Steele et al.,2011; Chow et al., 2014).ey found some expected associations between organisms suchas associations between cyanobacteria and cyanophage (Chow et al., 2014) and someassociations in the cyanobacterial communities suggestive of functional redundancy.e networks of viruses had dišerent properties as opposed to the bacterial and protistancommunities. ese studies examined T4-like myoviruses with bacteria and protists,however, these types of studies have not yet been done with other groups or viruses.1.7.2 Co-occurrence and co-evolution of viruses and microbes over timeIn a chemostat experiment Marston et al. (2012) examined the temporal dynamics of aSynechococcus sp. strain and one phage infecting it, and observed rapid diversicationof both hosts and viruses over time (167 days). In replicates they saw slightly dišer-ent amounts of diversication, but all hosts and viruses showed cycles of co-evolutionwith rapid changes and extinction of some genotypes. ere was also the diversica-tion of phenotypic traits such as increased infectivity and resistance. With evidence ofsuch rapid co-evolution of viruses and hosts over time, it is important to examine co-occurrence and change of genotypes over time. Examining these co-occurrences in anatural setting will determine if there are any other factors that drive the diversicationof these viruses and of their hosts. Furthermore, there was evidence for viral control ofbacteria diversity in Sandaa et al. (2009)’s mesocosm experiment. ey saw that thenumber of virus-hosts pairs was similar between mesocosms, but viral compositiondišered between mesocosms. is observation gives credence to the idea proposedalongside the “Killing theWinner” theory that the number of niches are limited for virus-hosts pairs and a dišerent mechanism controls the composition of the virus-host pairs(ingstad, 2000).14chapter 11.8 influence of environment on viral communitiese environment is an important driver of bacterial and eukaryotic communities overtime. With seasonality it is easily imaginable that these communities are driven byenvironmental factors. As well there may also be large biotic factors that drive thecommunity composition. ere are many environmental factors that inžuence the in-fectivity of viruses such as temperature, salinity, and nutrients (as reviewed in Mojicaand Brussaard (2014)).is occurs even though viruses are oŸen physically able to with-stand much greater ranges of environmental parameters than their hosts. For example,there are temperatures where certain viruses are better able to infect hosts comparedto other viruses (reviewed in Mojica and Brussaard, 2014; Kendrick et al., 2014). us,environmental parameters may be driving part of the diversity of the viral communityby ašecting the infectivity of the viruses.Viruses inžuence their host communities by forcing them into a co-evolutionaryarms race through selective pressure. Co-evolution, however, is not the only pressureon viral diversity and community composition. Changes in community compositionhave been associated with, but not limited to environmental parameters such as, salinity,temperature, chlorophyll a, the mineral jarosite, carbon, and nutrient žuxes (Short andSuttle, 2003; Kyle et al., 2008; Sandaa et al., 2009). Although some aquatic virusesmay beendemic to specic environments (Williamson et al., 2008), some genotypes appear to becosmopolitan with the most abundant types being widely distributed (Short and Suttle,2005; Angly et al., 2006). Combining data on compositional changes in specic groupsof viral families, potential host communities, and in abiotic parameters will enable thedetermination of overall and specic drivers of genotypic change in viral communities.15chapter 11.9 viral and microbial dynamics during ecologicaldisturbances1.9.1 What are ecological disturbances?Disturbances are ecological perturbations that occur on scales of either days-weeks, pulsedisturbances, and those that occur overweeks-months, press disturbances. Disturbancesof whole communities and specically themicrobes have been important to study from acommunity ecology standpoint since the changes in one population can inžuence otherpopulations and the overall community (Shade et al., 2012a,b; Banks et al., 2013). Onecommunity disturbance in which the viral communities are oŸen deeply implicated isduring phytoplankton blooms (Bratbak et al., 1996;Wilson et al., 2002a). Phytoplanktonblooms are events whereby the phytoplankton biomass increases very quickly and thereare high numbers of cells, oŸen from one species of phytoplankton. ese blooms canlast for days to months depending on the species and the system. ese blooms have alarge inžuence on the abiotic and biotic parameters.1.9.2 Viruses during bloomsViruses have oŸen been associated with the demise of blooms in mesocosms and innaturalwaters (Bratbak et al., 1996;Wilson et al., 2002a,b). For instance, viruses infectingEmiliania huxleyi have been found in high abundance at the termination of blooms.e viruses associated with the bloom can be composed of one or multiple genotypes(Higheld et al., 2014). It is unclear what causes the bloom to terminate with one orseveral viral genotypes. During a bloom of alga, Heterosigma akashiwo, Tarutani et al.(2000) saw the evolution of both viral andhost strains based on infectivity patterns. Overthe progression of the bloom there were dišerent host strains present and the infectivityof the host strains decreased over the course of the bloom.16chapter 11.9.3 Ešect of disturbance on other microbial communitiesese disturbance events, specically phytoplankton blooms, can have large ešects onthe bacterial communities as there are heterotrophic bacteria that can e›ciently andešectively use products from the phytoplankton (such as amino acids, carbohydrates,organic acids, polysaccharides, proteins, nucleic acids, and lipids) and thus quickly in-crease in abundance during a phytoplankton bloom (Buchan et al., 2014; Teeling et al.,2012).e bacteria that tend to respond to these events are predominantly the Flavobac-teria, Gammaproteobacteria and some of the Alphaproteobacteria.ese bacteria tendto be good at degrading phytoplankton products (Buchan et al., 2014). Interestingly, thebacterial communities tend to remain even during a bloom, with no one type dominat-ing the community (Delmont et al., 2014).us the evenness of the bacterial communitycan remain high even though the bacterial community composition can change drasti-cally throughout the bloom.1.10 choice of genetic targets1.10.1 Use of marker genes for studying microbial communitiesUnlike cellular organisms, viruses have no universal shared genes that can be used foroverall phylogeny like the well-conserved small subunit 16S ribosomal RNA (rRNA)gene for prokaryotes and the 18S rRNAgene for eukaryotes (Woese and Fox, 1977;Woeseet al., 1990). Nevertheless, using the polymerase chain reaction (PCR), several well-conserved viral marker genes (such as DNA pol (Chen and Suttle, 1995), major capsidprotein (Filée et al., 2005), and the RNA dependent RNA polymerase (Culley et al.,2003)) have been used to illuminate the genetic richness of dišerent groups of aquaticviruses. ese genes have highly conserved regions, which are used to determine thepresence of viral families in a sample. However, these genes also have variable regions,which determine the richness of a related group of viruses. ese conserved genes are17chapter 1good surrogates for whole genomes since the phylogeny of the conserved viral genes iscongruent with the phylogeny of the whole viral genomes (Filée et al., 2005).1.10.2 Metagenomics vs. marker genesViral metagenomics is the characterization of the viral community by sequencing thetotal viral nucleic acid in a sample. Metagenomics of viruses has revealed a huge amountof information about the kinds of viruses in the ocean (Breitbart et al., 2002; Angly et al.,2006; Culley et al., 2006; Hurwitz and Sullivan, 2013; Brum et al., 2015). is approachis very useful for examining unknown viruses and for discovering unknown functionsfromdišerent viruses since no previous knowledge of the communities is needed. Whileviral metagenomic data generally show high richness, many of the sequences have nosignicant homologues in Genbank. In coastal viral communities 65 % of the dsDNA(Breitbart et al., 2002), and 63-81 % of RNAmetagenomes had no signicant matches toGenbank (Culley et al., 2006). Additionally, since certain classes and families of virusesare only found infecting specic hosts and specic taxonomic host ranges (Koonin et al.,2008) the viral metagenomic approach could eventually be used as a proxy for determin-ing which hosts are present in the environment.In recent years this metagenomic approach was useful for characterizing the mostabundant viruses in communities. For communities with high richness, however, therarer viruses may be less well characterised. e amount of sequencing has increasedrapidly in the past few years coupled with decreased costs. If the present studies were tobe conducted today, the metagenomic approach could be appropriate for the questionsrelated to community diversity and structure. Metagenomics, at that time (circa 2010),provided a broad and shallow look at the viral communities compared to marker geneapproaches with high-throughput sequencing. Using targeted marker gene sequencingallows a deeper look into the composition, and structure of specic communities overtime. However, with continued increases in sequencing output, this argument is beingrevisited for these types of ecological questions (Sullivan, 2015).18chapter 11.10.3 Description and distribution of the T4-like myoviruses (Caudovirales) using themarker gene gp23Many dišerent groups of viruses have been found to be ecologically important andnumerous in the environment (Adriaenssens and Cowan, 2014). ese include virusesinfecting eukaryotes such as the Phycodnaviridae, and Picornavirales, and viruses infect-ing bacteria including Caudovirales, and theMicroviridae (Gokushovirinae). Two genemarkers were chosen that target viruses infecting two dišerent host domains. e rstgene marker targets a group of viruses infecting bacteria (bacteriophage). ere havebeen observations of bacteriophage in aquatic systems since the 1950’s (Spencer, 1955,1960). Many of the viruses that infect bacteria fall within the order Caudovirales.eseare double-strandedDNA (dsDNA) viruses whose genomes range in size from 33-244kb,and have a morphology of a head (viral capsid) and tail. Within the Caudovirales thereare 3 main groups: theMyoviridae with contractile tails, the Siphoviridae with long tails,and the Podoviridae with short tails.One diverse and well-studied group of bacteriophage in the marine environment istheT4-likemyoviruses. Filée et al. (2005) designed degenerate primers for gp23, the genewhich encodes themajor capsid protein in T4-likemyoviruses. Filée et al. (2005) found alarge diversity of environmental viruses that were distantly related to the cultured virusesin this group. Filée et al. (2005) found high genetic richness in 3 geographically sepa-rated viral samples from British Columbia, the eastern Gulf of Mexico and the westernArctic Ocean. Within this group there are many ecologically important viruses infect-ing cyanobacteria including those infecting the crucial marine species of cyanobacteriaSynechococcus and Procholorococcus. ese viruses can have a broad or specic hostrange (Sullivan et al., 2003). is group also includes some of the pelagiphage whichinfect a bacterial clade that includes some of the most abundant bacteria in the ocean,SAR11 (Zhao et al., 2013). Temporal studies have highlighted seasonal patterns in theT4-like myoviruses; furthermore some T4-like myoviruses are constant members of19chapter 1the community while others have a more ephemeral nature (Chow and Fuhrman, 2012;Pagarete et al., 2013; Needham et al., 2013).1.10.4 Studying aquatic RNA viruses using the marker gene RdRpAnother important group of viruses in the marine environment are a subset of the vi-ral order the Picornavirales. is order of viruses is a group of single-stranded RNAviruses that infect eukaryotes, specically plants, invertebrates and protists (LeGall et al.,2008; Tomaru et al., 2015).ey are small, icosahedral viruses with a capsid diameter of~25nm, no overlapping reading frames, a conserved RNA-dependent RNA polymerase,all RNAs are translated into a polyprotein before processing, and have a genome sizeof ~9kb (Le Gall et al., 2008). In the marine environment, these viruses are importantpathogens of protists (Tomaru et al., 2015), and include isolates from the viral familythe Marnaviridae, and the genera Bacillariornaviridae and Labyrnaviridae. Sequencesin marine viral RNA metagenomes have hit reference genomes from families withinthis order such as the Tombusviridae and the Dicistroviridae (Culley et al., 2006, 2014).e isolated viruses are pathogens of ecologically important marine species. e typevirus of theMarnaviridae is theHeterosigma akashiwoRNA virus (Tai et al., 2003) whichinfects a widespread alga,Heterosigma akashiwo, which can formmassive blooms in thecoastal ocean. Other isolated viruses infect two ecologically important chain-formingdiatoms that can be important in blooms in the coastal ocean: 1) theRhizosolenia setigeraRNA virus 01 infecting Rhizoselenia setigera, and 2) Chaetoceros tenuissimus RNA virus01, and Chaetoceros socialis f. radians RNA virus 01 which infect dišerent species ofChaetocerous sp..e nal known isolate in these marine picorna-like viruses is a virusisolated from a pennate diatom,Asterionellopsis glacialisRNAvirus (Tomaru et al., 2012).Using degenerate primers amplifying the RNA dependent RNA polymerase (RdRp)gene from positive sense single-stranded RNA picorna-like viruses (order Picornavi-rales), Culley et al. (2003) described a high richness of sequences from the Strait ofGeorgia (SOG), British Columbia which all had low identity to homologues in Genbank.20chapter 1Later, using redesigned picorna-like primers, Culley and Steward (2007) described 5 newputative genera and 24 new putative species of picorna-like RNA viruses from studysites in Hawaii and Monterrey Bay, California. ese studies illustrate the widespreadoccurrence and high diversity of picorna-like viruses in the ocean. Furthermore, thesecommunities ofmarine picornaviradswere highly uneven (Culley et al., 2006;Gustavsenet al., 2014). To examine the dynamics of these highly diverse communities, ampliconsequencing is useful for examining the ne-scale dynamics of these viruses, their related-ness, their putative hosts, and their associations with phytoplankton blooms since theyinfect known bloom-formers.1.10.5 Examining marker genes of bacteria and eukaryotes for richness and compositionViruses can only replicate within hosts and thus their hosts are of great interest duringany study of viral ecology. Within the past ten years there have been large developmentsin amplicon sequencing for examining themicrobes that could be hosts for the viral com-munities (Sogin et al., 2006; Lozupone and Knight, 2007; Logares et al., 2014). Bacterialcommunities are assessed using conserved and variable parts of the 16S rRNA gene.eprimer set developed by Baker et al. (2003) was used in this dissertation. It was chosenbecause the primer set was compatible with a 454 study that was on-going. When testedagainst the Silva database (Quast et al., 2013) the primer set had 87% coverage of thebacteria in the database. Similarly, for eukaryotic communities the 18S ribosomal RNAgene was used to examine the dynamics of the eukaryotic communities. A primer setdeveloped by Diez et al. (2001) was used to assess microeukaryote communities.e setwas used because the short product length was deemed to be a useful and e›cient wayto take advantage of the Miseq Illumina platform (which was used for sequence datageneration in Chapters 3, 4, and 5).21chapter 11.11 caveats (methodological considerations)1.11.1 Water massesis dissertation will discuss the temporal dynamics of marine viral, bacterial and eu-karyotic communities. However, it must be understood that the same water mass is notbeing followed through time.ese projects use a Eulerian sampling scheme where thewater that comes by a point is sampled over time rather than a Lagrangian samplingscheme – where water masses would be followed and sampled through time (Hewsonet al., 2006a; Gilbert et al., 2011; Fuhrman et al., 2015). Eulerian studies have shownimportant and repeatable dynamics of the marine waters (Gilbert et al., 2011).us eventhough it is not following the same water masses there are predictable patterns found inthese waters. Also, there is heterogeneity in seawater at the microscale level (Seymouret al., 2006), so the 60L of sub-surface water collected and used in these projects wouldlikely have containedmany small niches (distinct habitats), thatwere ešectively averagedduring sample collection.1.11.2 OTU clusteringAnother consideration is that individual sequence reads need to be grouped together at acertain similarity to examine these communities.e termOperational TaxonomicUnit(OTU) is used to describe these groupings of sequences. For bacteria and eukaryotesmany have used OTU sequence similarity of 97 % for putative species (Stackebrandtand Goebel, 1994), although it is arbitrary and can dišer for dišerent species of bacteria(Koeppel and Wu, 2013b). New approaches, such as oligotyping (used in Chapter 5)provide complementary approaches to OTU-based methods by analysing the sequencereads without set similarity cut-ošs.22chapter 11.11.3 Taxonomic classicationTaxonomic classication of a sequence is only as good as the database used. Althoughshort sequences of partial genes can be surprisingly ešective at identifying organisms,they can also have lower resolution than full-length sequences of the target genes. Bac-terial and eukaryotic sequences were classied using the Silva database (Quast et al.,2013). ese reference databases are reliant on isolated organisms and can be limitedfor sequences lacking cultured representatives. Furthermore, viral databases are toosparse (since there is a paucity of isolated marine viruses) to be able to classify viruses.Unknown viral sequences can be examined for patterns by their occurrence in samplesand in the context of potential hosts. is dissertation instead built phylogenies toexamine the viruses in a genetic context.1.11.4 Problems associated with PCR and sequencingAll of the marker genes examined in this dissertation use PCR to amplify the targets.Although this is a widely-used technique (and goes bymany names such as environmen-tal sequencing, eDNA, amplicon deep sequencing, tag sequencing, pyrotags) there aresome caveats associated with it.ere is known bias associated with PCR amplicationwhere the amount of template can skew the amplication. Making sure to use appro-priate amounts of template can help mitigate this bias (Lee et al., 2012). Another biasis in primer mismatch where dišerent species can be preferentially amplied based ontheir sequence or structure (Acinas et al., 2005; Sipos et al., 2007) and this is especiallyimportant with the use of degenerate primers. Furthermore, sometimes with limiteddatabases primers can be designed which might miss relevant and important speciesin an environment. Such is the case with the Earth Microbiome primers which missedmany important bacterial species (Apprill et al., 2015; Parada et al., 2016). As alluded toin the previous section this can be mitigated by checking primers against databases tosee what would be amplied.23chapter 1In this dissertation two dišerent kinds of sequencing were used, 454 Titanium andIllumina Miseq. Both are both high-throughput methods of sequencing that are fre-quently used for community analysis of microbial communities because of their highamounts of sequence reads returned. As previously discussed, high-throughput sequenc-ing ošers a resolution and sensitivity not approached by (Automated) Ribosomal Inter-genic Spacer Analysis (ARISA), T-RFLP and DGGE. However, both 454 Titanium andIllumina Miseq sequencing come with their own caveats.e 454 sequencing approach,used in Chapter 2, has the problem of inaccurate calling of homopolymers resulting ininaccurate strings of one base (Quince et al., 2009) which tended to cause inžated levelsof diversity (Kunin et al., 2009). To reduce this potential bias, denoising (Quince et al.,2009; Reeder and Knight, 2010) was used to minimize some of the errors associatedwith 454 sequencing. Denoising attempts to minimize the loss of sequence informationby using the žowgrams from the 454 sequencer to distinguish erroneous sequences fromreal sequences.In Chapters 3, 4, and 5 Miseq Illumina sequencing was used. Miseq had the lowesterror rate (0.1 substitutions per 100 bases) and highest throughput of any of the “bench-top” sequencers at the time (Loman et al., 2012). However there are still errors associatedwith it (Schirmer et al., 2015). Ways to minimize the per-base error rate are by lteringsequences to maintain only highest quality sequences (e.g. above a quality score of 20which represents 1% error), trimming low quality ends and sequencing targets shortenough for paired-end sequence reads to overlap, thus decreasing the probabilities forerrors (Bokulich et al., 2013).For microbial ecology studies it must be assured that the samples are comparableby normalizing the sampling/sequencing ešort. In ecology, one of the few agreed uponideas approaching a law is that of the species-area or ešort curves (Rosenzweig, 1995),e.g. the more you sample the greater richness you see. e same is oŸen true withsequencing where the more you sequence the greater the observed richness until youreach an asymptote. us, because of stochasticity in the sequencing process and the24chapter 1need to have samples on dišerent sequencing runs, the data need to be normalized tobe able enable comparison among samples. A common approach for normalizing readscalled “rarefying”, is to use the number of reads in the smallest library and randomlypick that number of reads (without replacement) from the other libraries (collection ofsequence reads from one sample).is method of examining microbial and viral communities can be considered semi-quantitative and thus able to compare increases and decreases in samples, not absolutenumbers. Mock communities, where known amounts of genomic DNA are used withthe same protocols, have been used to help improve and validate the use of these tech-niques (Schloss et al., 2011). Wherever possible these techniques have been incorporatedinto the studies (e.g. using cloned mock communities to assess error rates and to helpassess appropriate percent identity for clustering of sequence reads).1.12 research objectives and outline of thesis1.12.1 Overall goals of thesise overarching goals of this dissertation were to document temporal changes in the ge-netic composition of natural assemblages of marine viruses, and to relate these changesto variations in the physical and biological environment, including the composition ofpotential host communities. Viral gene markers examined at temporal and spatial timescales will examine whether genotypes abundant in one sample remain abundant overtime or if they remain, but become rarer or undetectable.Overall approach:• Collect samples from coastal site Jericho Pier every 2 weeks for one year. Addi-tionally sample Jericho Pier every other day for a tidal cycle in the winter and thesummer.25chapter 1• Quantify richness in viral samples by amplicon sequencing of T4-like myoviruses(gp23), and marine picorna-like viruses (RdRp), and in putative host samples ofbacteria (16S rRNA gene) and eukaryotes (18S rRNA gene) amplicons.• Compare communities for compositional overlap, community richness and diver-sity.• Determine the drivers of viral diversity and community composition over timefrom measured environmental parameters to explain how viral diversity is main-tained in dišerent viral groups.1.12.2 Description of chaptersChapter 2: High temporal and spatial diversity inmarineRNA viruses impliesthat they have an important role in mortality and structuring planktoncommunities — Ecological questions about the distribution of marine viruses overtime and space have been examined more extensively in bacteriophages, particularlythose infecting cyanobacteria.Hypotheses: If the dynamics of marine bacteriophages and marine RNA viruses aresimilar, some RNA viral taxa will persist temporally and spatially, while other taxa willbe detected sporadically.Approach: To test this hypothesis, two samples, taken ve months apart at the samelocation, and three samples taken within hours of each other, but 20 km apart in thesame coastal basin were examined using 454 Titanium sequencing of the viral markergene: RNA dependent RNA polymerase (RdRp). Phylogeny and relative abundance ofOTUs in the community were used to characterize the temporal and spatial dišerencesamong the samples.Chapter 3: Examination ofmarine viral community structure reveals phylo-genetic shifts over time — e viral gene markers were used to examine high res-olution temporal dynamics, phylogenetic distribution over time, and putative changes26chapter 1in response to changes in composition to the host communities and to disturbances inthe environment.Hypotheses: 1) Related viruses should share similar ecology and thus should showthe same occurrence patterns over time.e structure and composition of viral commu-nities has been described to follow seed bank type distributionwhere there aremany rareviruses and few abundant ones. 2)ere would be shuœing of individual viral OTUs inthe communities, but communities would remainmostly dominated by a few viral typesover time. 3)e phylogenetic structure of viral communities would be inžuenced bythe phylogenetic structure of their putative hosts andwould show temporal shiŸs relatedto the seasonal progression of their hosts.Approach: Using a one-year time series, sampled every two weeks at Jericho Pier allof the marker genes for viruses and potential hosts were analysed phylogenetically andthe community structure was examined.Chapter 4: Network analysis of Jericho Pier microbial time series — Virusespresent at the site must be infecting a host that is also present and thus it would beexpected that viruses co-occur with their specic host. e environment has a role indriving the diversity of these viruses beyond just driving the diversity of hosts sinceenvironmental parameters can inžuence the infectivity of viruses.Hypotheses: At least co-occurrence (with a potential time lag) of viral OTUs witha putative host and potentially others would be expected. Within networks composedof the two groups of viruses associated to their putative hosts it would be expected toobserve dišerent patterns of co-occurrence since the viral host ranges and life historiesare dišerent. It is hypothesized that that in addition to the inžuence of host communities,viral communities will also be driven by environmental parameters.Approach: Using data from the year-long study at Jericho Pier to form local similar-ity analysis networks to determine whichOTUs co-occur strongly and also to determineif there are time-lags associated with the patterns of strongest co-occurrence of the pair-27chapter 1wise relationships. Use redundancy analysis and variation partitioning to determine theenvironmental and biotic drivers of diversity and community composition over time.Chapter 5: Viral and heterotrophic protistan control of a phytoplanktonbloom in coastalwaters — Hypotheses: Considering the strain level dišerences ininfectivity seen during phytoplankton blooms and in overall temporal dynamics of mi-crobial communities it was predicted that during a bloom of eukaryotic phytoplanktonthere would be a progression of strain level OTUs of themain bloom former attributableto specic strain level viral pressure.Approach: Examine how viral, bacterial and eukaryotic richness change during abloom of one eukaryotic species which occurred during a high-resolution time seriessegment of the Jericho Pier sampling. Use Shannon entropy decomposition (oligotyp-ing) to sub-divide OTUs into ner resolution to determine how the single OTUs thatdominate the communities during the bloom evolve and how they compare to the restof the time series.1.13 significanceese studies provide a detailed genotypic examination of temporal dynamics in thecomposition of DNA and RNA viral communities and their putative hosts. Notably, thisis the rst study that uses all of these communities together while using high throughputsequencing. is dissertation will advance the eld in several important areas: addi-tional evidence for community-level theories related to environmental viruses such asthe Bank theory and Killing theWinner, insight into dynamics of the phylogenetic struc-ture of viral communities, the dynamics of ephemeral compared to persistent OTUs,the determination of which environmental parameters play an important role in theco-occurrence of viruses and hosts, and examination of the ešect of disturbances onmicrobial communities.28chapter 2High temporal and spatial diversity in marine RNAviruses implies that they have an important role inmortality and structuring plankton communities12.1 summaryViruses in the orderPicornavirales infect eukaryotes, and arewidely distributed in coastalwaters. Amplicon deep-sequencing of the RNA dependent RNA polymerase (RdRp)revealed diverse and highly uneven communities of picorna-like viruses in the coastalwaters of BritishColumbia (B.C.), Canada. Almost 300000pyrosequence reads revealed145 operational taxonomic units (OTUs) based on 95% sequence similarity at the amino-acid level. Each sample had between 24 and 71 OTUs and there was little overlap amongsamples. Phylogenetic analysis revealed that some clades of OTUs were only found atone site, whereas, other groups included OTUs from all sites. Since most of these OTUsare likely from viruses that infect eukaryotic phytoplankton, and viral isolates infectingphytoplankton are strain-specic; each OTU probably arose from the lysis of a specicphytoplankton taxon. Moreover, the patchiness in OTU distribution implies continu-ous infection and lysis by RNA viruses of a diverse array of eukaryotic phytoplanktontaxa. Hence, these viruses are likely important elements structuring the phytoplanktoncommunity, and play a signicant role in nutrient cycling and energy transfer.1Chapter 2 has been previously published as: Gustavsen, Julia Anne, Danielle MWinget, Xi Tian, andCurtis A Suttle. High Temporal and Spatial Diversity in Marine RNA Viruses Impliesatey Havean Important Role in Mortality and Structuring Plankton Communities. Frontiers in Microbiology 5, no.703 (2014).29chapter 22.2 introductionViruses are highly abundant and widespread in the oceans (Bergh et al., 1989; Suttle,2005). Beyond their impacts on host mortality, viruses are signicant mediators ofbiogeochemical processes, horizontal gene transfer, and host community diversity inthe oceans (Fuhrman, 1999; Wilhelm and Suttle, 1999; Suttle, 2005). Marine viruses areimportant pathogens of phytoplankton (Brussaard, 2004a) and have been implicatedin the termination of blooms (Schroeder et al., 2003; Nagasaki et al., 1994) and withsuccession in phytoplankton communities (Mühling et al., 2005). Viruses have beencharacterized that infect a wide variety of phytoplankton such as haptophytes (Brat-bak et al., 1993), prasinophytes (Derelle et al., 2008; Mayer and Taylor, 1979; Brussaardet al., 2004), chlorophytes (Van Etten et al., 1981), diatoms (Shirai et al., 2008), anddinožagellates (Tomaru et al., 2004). Viruses infecting eukaryotic phytoplankton gen-erally have very narrow host ranges (Short, 2012). Viruses infecting marine phytoplank-ton have genomes comprised of double-stranded DNA (dsDNA), single-stranded DNA(ssDNA), double-stranded RNA (dsRNA), and single-stranded RNA (ssRNA) (as re-viewed in Short, 2012). eir genomes and particle sizes range from very large dsDNAviruses in the Phycodnaviridae to very small ssDNA and ssRNA viruses belonging to thegenus Bacilladnavirus and order Picornavirales, respectively.e order Picornavirales iscomprised of positive-sense, ssRNA viruses that infect eukaryotes (Le Gall et al., 2008),including ecologically important marine protists. ese viruses are small (25-35nm),icosahedral, and have a conserved genomic organization that includes a replication areacomprised of a type III helicase, a 3C-like proteinase, and a type I RNA dependent RNApolymerase (Sanfaçon et al., 2009). Isolates in the Picornavirales that are pathogens ofmarine protists infect a wide diversity of hosts including the bloom-forming raphido-phyte Heterosigma akashiwo (Tai et al., 2003) (viral family Marnaviridae), the thraus-tochytrid Aurantiochytrium sp. (Takao et al., 2005; Yokoyama and Honda, 2007) (vi-ral genus Labyrnavirus) and the cosmopolitan diatoms Rhizosolenia setigera (Nagasakiet al., 2004) and Chaetoceros socialis (Tomaru et al., 2009) (viral genus Bacillarnavirus).30chapter 2Viruses in the Picornavirales appear to be common and widely distributed in coastalwaters (Culley et al., 2003; Culley and Steward, 2007).Metagenomic and targeted gene studies are uncovering the diversity of marine RNAviruses. For example, phylogenetic analysis of RNA-dependent RNApolymerase (RdRp)sequences from seawater samples supports a monophyletic marine group within thePicornavirales (Culley et al., 2003; Culley and Steward, 2007; Tomaru et al., 2009; Culleyet al., 2014) and several divergent clades within this marine group (Culley et al., 2003;Culley and Steward, 2007; Culley et al., 2014). Additionally, metagenomic analyses revealthat there are numerous sequences from aquatic RNA viruses that cannot be assignedto known taxa (Culley et al., 2006; Djikeng et al., 2009; Steward et al., 2012; Culleyet al., 2014). Despite the high diversity of marine RNA viruses (Lang et al., 2009), thespatial and temporal distribution of dišerent phylogenetic groups remains unreported,although there is evidence that the taxonomic structure of marine RNA viral commu-nities is highly uneven. For example, in one sample from a metagenomic study fromthe coastal waters of British Columbia, 59% of the reads assembled into a single contig,while in a second sample 66% of the reads fell into four contigs, with most falling intotwo genotypes (Culley et al., 2006). However, with only a few hundred reads in totalfrom the two samples, the coverage of the communities was low. Similarly, RNA viralmetagenomic data from a freshwater lake (Djikeng et al., 2009) showed little identicalsequence overlap among communities, although there was broad taxonomic similarityover time within a location.Ecological questions about the distribution of marine viruses over time and spacehave been examined more extensively in bacteriophages, particularly those infectingcyanobacteria. For example, somedata reveal no clear patterns of biogeography in cyanophageisolates locally (Clasen et al., 2013), regionally (Jameson et al., 2011) or more globally(Huang et al., 2010). Other data have shown patterns at a regional scale (Marston et al.,2013), where communities in basins that were connected were most similar and thosethat were separated by land or current boundaries were the least similar. Other data31chapter 2for marine bacteriophages have shown temporal variability (Chow and Fuhrman, 2012;Clasen et al., 2013; Marston et al., 2013; Chen et al., 2009; Wang et al., 2011). If thedynamics ofmarine bacteriophages andmarineRNAviruses are similar, someRNAviraltaxa will persist temporally and spatially, while other taxa will be detected sporadically.To test this hypothesis, we examined two samples, taken ve months apart at the samelocation, and three samples taken within hours of each other, but 20 km apart in thesame coastal basin.We used high-throughput 454 pyrosequencing to obtain deep coverage of RdRpamplicon sequences and compare the richness of viruses in the Picornavirales amongsamples from the coastal waters of British Columbia, Canada. e results revealed aphylogenetically diverse and spatially variable community of viruses, suggesting thattaxon-specic lytic events are important in shaping the phytoplankton community.2.3 materials and methods2.3.1 Sampling locationsTo assess viral communities from dišerent coastal habitats, we collected samples fromthree sites in the Strait of Georgia (49○ 14.926N 123○ 35.682W, 49○ 17.890N 123○ 43.650W,and 49○ 23.890N 123○ 59.706W), and from Jericho Pier (49○ 16’36.73N, 123○ 12’05.41W)in British Columbia, Canada (Figure 2.1). e Strait of Georgia (SOG) is an estuarine-inžuenced basin that is on average 22 km across, 222 km long and 150 m deep. eupper 50m of SOG is where most of the variability in physical and chemical parametersoccurs. Jericho Pier (JP) is adjacent to the shoreline, in a well-mixed locationwithmixedsemidiurnal tides.2.3.2 Sample collectionOn 28 July 2010, a rosette equipped with a Seabird SBE 25 CTD (equipped with SeabirdSBE 43 dissolved oxygen sensor, WET Labs WETStar žuorometer, WET Labs C-Star32chapter 2Figure 2.1: Location of sampling sites. Map showing the location of sampling siteswithin the Strait of Georgia (SOG) and Jericho Pier, adjacent to Vancouver, BritishColumbia, Canada. Jericho Pier was sampled in summer (JP-S) and fall (JP-F).33chapter 2transmissometer, Biospherical InstrumentsQSP2200PDPAR sensor) andGeneralOcean-ics GO-FLO bottles was used to collect 12 L of water from ve depths between 2 and 16m at each of three stations in SOG.e ve depths from each station were combined toobtain three integratedwater samples.e depths were selected to attempt to encompassthe viral diversity at each station, and consisted of a near-surface sample, a sample fromthe isothermal zone below the mixed layer, and three samples spanning the chlorophyllmaximum.At JP, 60 L of water was pumped from the 1-m depth on 10 July and 12 October 2010.Salinity and temperature at JP were measured using a YSI probe (Yellow Springs, Ohio,USA). For all samples, the water was ltered through 142-mmdiameter, 1.2-µmnominalpore-size glass-ber (GC50 Advantec MFS, Dublin, CA., USA) and 0.22-µm pore-sizepolyvinyldine (Millipore Bedford, MA, USA) lters.e viral size fraction in the ltratewas concentrated to ~500 mL (viral concentrate) using tangential žow ultraltrationusing a 30kDa MW prep-scale Spiral Wound TFF-6 cartridge (Millipore) (Suttle et al.,1991). Phosphate, silicate and nitrate+nitrite concentrations were determined in dupli-cate 15-mL seawater samples ltered through 0.45-µm pore-size HA lters (Millipore)and stored at -20○C until air-segmented continuous-žow analysis on a AutoAnalyzer3 (Bran & Luebbe, Norderstedt Germany). Chlorophyll a (Chl a) was determined intriplicate by ltering 100 mL of seawater onto 0.45 µm pore-size HA lters (Millipore),and storing the lters in the dark at -20○C until acetone extraction and then analysedžuorometrically (Parsons et al., 1984). e average and standard error of the replicateswas calculated for each sample.2.3.3 Nucleic-acid extraction and PCRe viral concentrate was ltered twice through 0.22-µm pore-size Durapore PVDFlters (Millipore) in a sterile Sterivex lter unit (Millipore).e ltrate, containing virus-sized particles, was pelleted by ultracentrifugation (Beckman-Coulter, Brea, California,USA) in a SW40 rotor at 108 000 g for 5 h at 12○C.e pellet was resuspended overnight34chapter 2in 100 µL of supernatant at 4○C. To digest free DNA, the pellets were incubated with1U/µl DNAse with a nal concentration 5 mM MgCl2 for 3 h at room temperature.Nucleic acids were extracted using a Qiamp Viral Minelute spin kit (Qiagen, Hilden,Germany) according to the manufacturer’s directions. To remove DNA, the extractedviral pellets were digested with DNase 1 (amplication grade) (Invitrogen, Carlsbad,California, USA) and the reactionwas terminated by adding 2.5mMEDTA(nal concen-tration) and incubating for 10min at 65○C.ComplementaryDNA (cDNA)was generatedusing Superscript III reverse transcriptase (Invitrogen)with randomhexamers (50ng/µl)as per themanufacturer. PCRwas performedusing primer setMPL-2 for a targeted set ofthe marine picornavirus-like RdRp (Culley and Steward, 2007). Each reaction mixture(nal volume, 50 µl) consisted of 50 ng of cDNA, 1x (nal concentration) PCR bušer (In-vitrogen), 2mMMgCl2, 0.2mMof each deoxynucleoside triphosphate (Bioline, London,UK), 1 µMof each primer, and 1U PlatinumTaqDNApolymerase.e reactionwas runin a PCR Express thermocycler (Hybaid, Ashford, UK) with the following conditions:94○C for 75 s, followed by 40 cycles of denaturation at 94○C for 45 s, annealing at 43○C for45 s, and extension at 72○C for 60 s and a nal extension step of 9 min at 72○C. Negativecontrols were run with every PCR performed. PCR products were cleaned using theMinelute PCR purication kit (Qiagen).2.3.4 Library prep and pyrosequencingLibraries for each site were prepared for sequencing using NEBNext DNA Library Prepfor 454 kit (New England Biolabs, Ipswich, Massachusetts, USA) following the manufac-turer’s directions, and using Ampure beads (Beckman-Coulter) for size selection andpurication using a bead ratio of 0.8:1 beads:library. PCR amplicons were barcodedand sent for 454 Titanium pyrosequencing (Roche, Basel, Switzerland) at the GénomeQuébec Innovation Centre at the McGill University (Montreal, QC, Canada).35chapter 22.3.5 Sequence analysisSequencing reads were quality trimmed using length settings between 100 and 600 bp,with a maximum of 3 primer mismatches for the specic primer, and denoised usingthe denoiser algorithm in QIIME (version 1.7) with default settings for Titanium data(Caporaso et al., 2010; Reeder and Knight, 2010). Sequences were checked for chimerasusing UCHIME (Edgar, 2010) against a nucleotide database of RdRp sequences builtusing NCBI-BLAST+ by retrieving nucleotide sequences of the RdRp from Picornavi-rales viral isolates (accessed: 8 August 2012), and also using the denovo chimera checkin UCHIME.e overlap of results from these two methods was dened as chimeric se-quences, although none were found in this study. Non-chimeric sequences were queriedusing BLASTx (Altschul et al., 1990) with an e-value of 1e-3 against the database of RdRpviral isolates. All sequences with hits were retained and all sequences with no hits werethen queried against the non-redundant (nr) Genbank database (Benson et al., 2007)using BLASTx with an e-value of 1e-5. All the sequences identied as contaminantsor as unknown with only 1 read were removed. Remaining sequences were translatedto amino acids using FragGeneScan with the 454_10 training option (Rho et al., 2010).Sequences were grouped into operational taxonomic units (OTUs) using UCLUST at arange of similarities from 50% to 100% (Figure S 1) using the original seed sequences(centroids) as the output (Edgar, 2010). A similarity of 95% was chosen for this analysisfor the following two main reasons: 1) When the NCBI conserved domain alignmentfor the RdRp region (all Picornavirales) was analyzed for percent similarity, the onlysequences that displayed greater than 95% similarity in this region were strains of thesame virus. 2)e sequences from the control libraries (consisting of 1 clone) clusteredinto 1 sequence at this percentage (see Supplemental methods). us clustering at 95%similarity was a way to use biological and sequence-based information to inform ourchoice of cut-oš to collapse strain level variation and as a conservative approach to avoidvariation that may be present because of the sequencing platform. Control sequencesobtained by cloning and Sanger sequencing (see Supplemental Methods) were used to36chapter 2verify the sequence processing methodology. Raw and processed sequence data weredeposited in the NCBI BioProject database ID: PRJNA267690.2.3.6 Phylogenetic analysisAll OTUs with less than 5 reads were removed and the remaining OTUs were aligned us-ing prole alignment inMuscle (Edgar, 2010) to seed alignments of viral RdRps from theNCBIConservedDomainDatabase (Marchler-Bauer et al., 2010). Sequences fromotherenvironmental surveys were clustered in the same manner as the reads in this study (us-ingUCLUSTat 95%).e clusters are cluster number followed by theGenbank accessionnumbers contained in that cluster. 0: 33520549, 33520547, 33520541, 33520533, 33520527,33520521, 33520519, 33520517, 33520515, 33520513, 33520511, 33520509; 1: 157280772; 2 :33520525; 3: 157280768; 4: 157280770; 5: 568801536, 568801534, 568801530, 568801528,568801510; 6: 157280786; 7: 157280774; 8: 157280780; 9: 157280788; 10: 568801494, 568801492,568801488, 568801482, 568801480, 568801474, 568801470, 568801466, 568801464, 568801462,568801458; 11: 157280776; 12: 157280778; 13: 568801516; 14: 568801616, 568801614, 568801606,568801588, 568801586, 568801584, 568801582, 568801580, 568801574, 568801572, 568801564,568801550, 568801548, 568801540, 568801532, 568801522, 568801520, 157280784; 15: 568801508;16: 568801542, 568801538, 568801518, 568801486, 568801484, 568801478; 17: 568801612,568801610, 568801608, 568801604, 568801602, 568801600, 568801598, 568801596, 568801594,568801592, 568801590, 568801578, 568801570, 568801568, 568801552, 157280782; 18: 568801576,568801566, 568801546, 568801544; 19: 568801562, 568801556; 20: 568801526, 568801524,568801514, 568801512, 568801506, 568801504, 568801502, 568801500, 568801498, 568801496,568801490, 568801472, 568801460; 21: 568801476, 568801468; 22: 568801560, 568801558,568801554; 23: 157280744; 24: 157280758; 25: 157280748; 26: 157280746; 27: 157280742;28: 157280766; 29: 157280756; 30: 157280762, 157280754, 157280752; 31: 157280750; 32:157280760; 33: 157280764; 34: 33520545, 33520543, 33520535, 33520531, 33520529; 35: 33520539,33520537, 33520523, 33520507. Alignments were masked using trimAl with the automaticheuristic (Capella-Gutierrez et al., 2009) and edited manually. ProtTest 3.2 was used for37chapter 2amino-acidmodel selection (Darriba et al., 2011) before building the initial phylogenetictree using FastTree (Price et al., 2010). Final maximum likelihood trees were done withRAxML using sequences belonging to viruses in the Sequiviridae as the outgroup, andthe BLOSUM62 amino-acid model with 100 bootstraps (Stamatakis et al., 2008). etree was visualized in R (R Core Team, 2014) using the ape package (Paradis, 2012) andedited in Figtree (Rambaut, 2014).2.3.7 Statistical analysisGeneration of rarefaction curves by random resampling of OTU abundances was per-formed using the vegan package (Oksanen et al., 2013) in R (R Core Team, 2014). Rel-ative abundances were normalized by randomly resampling 10 000 times using vegan,normalizing to the library with the lowest number of reads and then taking the median.Rank-abundance curves were generated with ggplot2 (Wickham, 2009) using the OTUsper site normalized by the library with the lowest number of reads by taking the medianof 1000 random rarecations of theOTUabundance data. Scripts used in this project areavailable as part of QIIME and custom user scripts used to process the data are availableon github (Gustavsen, 2015).2.4 results2.4.1 Environmental parameterse environmental parameters ranged widely among samples (Figure 2.2). Chlorophylla values were lowest at Jericho Pier Fall (JP-F) at 0.16 µg L-1 (± 0.04) and highest at theStrait of Georgia Station 2 (SOG-2) at 3.2 µg L-1 (± 0.4). Silicate values for all sampleswere similar (range of 25.5 to 37.8 µM), except for Jericho Pier Summer (JP-S) whensilicate was lower at 6.2 µM (± 0.01). Phosphate ranged between 0.90 and 1.3 µM atJP-F, SOG-1, and SOG-4, but was lower at SOG-2 (0.58 µM (± 0.03)) and lowest at JP-S(0.06 µM (± 0.04)). Nitrate + nitrite values were more variable than the other nutrients38chapter 2and ranged from 1.35 µM (± 0.75) at JP-S to 14.6 µM (± 0.04) at JP-F.e SOG siteswere highly stratiedwith SOG-4 being themost stratiedwith a calculatedmixed-layerdepth of 2 m, while SOG-1 and SOG-2 were similar with a mixed-layer depth of 6 m(Table 2.1).Table 2.1: Description of samples and resulting sequencing information. * aŸer qualityltering and matching to the RdRp primer set.Location ofsamplingDate ofsamplecollection Latitude, Longitude Reads*Mixedlayerdepth(m)Total reads 300180Jericho PierSummer10 July 2010 49○ 16’36.73N, 123○12’05.41W74096 -Jericho Pier Fall 12 October201049○ 16’36.73N, 123○12’05.41W84907 -SOG 1 28 July 2010 49○ 14.926N, 123○35.682W55197 6SOG 2 28 July 2010 49○ 17.890N, 123○43.650W12269 6SOG 4 28 July 2010 49○ 23.890N, 123○59.706W73044 22.4.2 Analysis of RdRp sequencesAŸer quality ltering to remove homopolymers and contaminating reads, 300 180 readswere recovered from the 5 libraries of RdRp amplicons. At all sites the rarefaction curvesplateaued indicating that the depth of sampling was adequate to assess the communities(Figure 2.3). From these reads, 265 unique OTUs (at 95% similarity) were identied,39chapter 2Figure 2.2: Environmental parameters. A) Chlorophyll a with standard error of themean from triplicates. B) Silicate with standard error of the mean from duplicates. C)Phosphate with standard error of the mean from duplicates. D) Nitrate+ nitrite withstandard error of the mean from duplicates. E) Temperature with each point as one GO-FLO bottle (SOG samples) or from total seawater sample (Jericho samples). F) Salinitywith each point representing a seawater sample as one GO-FLO bottle (SOG samples)or from total seawater sample (Jericho samples).40chapter 2including 108 singletons. For further analysis OTUs were excluded that did not containrecognizable RdRp motifs (Koonin, 1991; Le Gall et al., 2008), generally did not alignwell with other RdRp sequences and those that were not present in any sample aŸernormalization. Using the above criteria there were 145 OTUs identied in all samples,and between 24 to 71 OTUs per site. e Jericho Pier samples had 116 OTUs of whichonly 10 (8.6%) were shared between sampling times (Figure 2.4). JP-S had the highestrichnesswith 71OTUs, of which 59 (83%)were unique, while JP-F had the second highestrichness (49 OTUs), of which 45 (92%) were unique. e SOG sites together had 64OTUs, none of which were shared among all sites. SOG-1 and SOG-2 had the lowestnumber of OTUs (24). SOG-1 had only three OTUs which were unique. However, 21(33%) were shared between SOG-1 and SOG-4, and 6 (9%) between SOG-2 and SOG-4.e majority of OTUs (75%) from SOG-2 were unique, whereas most OTUs from theother SOG sites were shared with other sites (87% for SOG-1 and 63% for SOG-4).Rank abundance curves of the viral OTUs showed that at each site most sequenceswere assigned to only a few OTUs (Figure 2.5). JP-S had the highest richness but theshallowest slope of these curves, demonstrating more evenness in the abundance ofOTUs than at the other sites. SOG-4 and JP-S had similar rank abundance curves thatwere much shallower that those of SOG-1 and SOG-2 (Figure 2.5).e OTUs that were observed in more than 5 reads were placed in phylogenetic con-text using a maximum likelihood RAxML tree (Stamatakis et al., 2008) with sequencesfrom previous RdRp gene surveys and isolated viruses (Figure 2.6). OTUs from thisstudy fell within a well-supported clade that includes all themarine isolates belonging tothe Picornavirales. Within this group there was a well-supported divide between OTUsgrouping in the Marnaviridae clade and those grouping with sequences from virusesinfecting diatoms and a thraustochytrid.e overall tree topology is not well supported,although there are a number of well-supported clades containing OTUs from this studyand other environmental sequences. e Marnaviridae clade had the greatest numberof OTUs (10) associated with it; whereas, very few OTUs (only OTUs 89, 107, 75 and41chapter 2Figure 2.3: Rarefaction curves. Rarefaction analysis of RdRp amplicons based on Chao1richness analysis of operational taxonomic units (OTUs) at 95% similarity. Rarefactioncurves were resampled using number of reads recovered per library. Rarefaction curvesplateau indicating adequate sequencing for these samples42chapter 2Figure 2.4: Euler diagrams of normalized RdRp OTUs. A) Euler diagram of JerichoPier samples. B) Euler diagram of SOG samples.e OTUs presented were from readsclustered at 95% similarity, comprise only OTUs that could be aligned to the NCBI CDDRdRp alignment, and contained the RdRpmotif C.e diagramswere constructed usingthe venneuler() algorithm (Wilkinson, 2011). e size of the circles is approximatelyproportional to the number of OTUs recovered per site. e overlap in the diagramdescribes OTUs that were found atmultiple sites and the non-overlapping areas describeOTUs that were unique to that site.43chapter 2Figure 2.5: Rank abundance by site. Relative abundance of OTUs in each sampleordered by rank abundance. OTUswere clustered at 95% amino acid similarity andOTUrelative abundances were normalized to the sample with the lowest number of reads44chapter 2120) from this study were assigned to clades primarily fromHawaii (Culley and Steward,2007; Culley et al., 2014). No clade contained OTUs from all sites. e Jericho Piersamples were the most phylogenetically diverse (Figure 2.6, Table 2.2), and containedOTUs (e.g. OTUs 6, 7, 35, 31, 84, 47, 14, 4, 34, 39, 44, 23, 8, 20) that fell into clades thatdid not contain OTUs from any of the SOG samples. Some clades contained OTUs fromboth JP-S and JP-F samples; however, many OTUs within the clades were unique to oneJericho Pier sample. Phylogenetic diversity dišered among samples, except for OTUsfrom the well-mixed SOG sites, some of which were present in dišerent clades resultingin similar phylogenetic diversity (Table 2.2).Table 2.2: Phylogenetic diversity, species richness. Phylogenetic diversity (PD) iscalculated as in (Faith, 1992). OTUs were the same as used in the construction of thephylogenetic tree and must have included ve or more reads.Phylogenetic diversity Species richnessJericho Pier Summer 16.25 46Jericho Pier Fall 13.62 30SOG Station 1 3.89 9SOG Station 2 3.26 4SOG Station 4 8.35 24e Strait of Georgia (SOG) sites were sampled within hours of each other, andthe water at each site was pooled from multiple depths above, below, and across thechlorophyll maximum. One of the most striking dišerences among sites was that SOG-1 and SOG -2 had mixed layer depths of 6m; whereas SOG-4 had a mixed-layer depthof 2 m, and much higher richness and phylogenetic diversity. Sites SOG-1 and SOG-2had the lowest phylogenetic diversity (Table 2.2). All the OTUs found at SOG-1 (33, 12,16, 9, 10, 27, 29) were within theMarnaviridae clade; similarly, all OTUs (5, 82, 2, 1) fromSOG-2 were within one distantly related clade. In both cases OTUs from these clades45chapter 2Figure 2.6: Phylogenetic tree with heatmap. Maximum likelihood (RAxML) tree ofRdRp OTUs that contained more than 5 reads and relevant Picornavirales sequences.Bootstrap values above 65% are labelled. Adjacent to the tips of the tree is a heatmapdisplaying the relative abundance of each OTU at 95% similarity by site. OTU relativeabundances were normalized to the sample with the lowest number of reads.46chapter 2occurred at SOG-4. SOG-2 did not have the high numbers of HaRNAV-related virusesthat were found in all other samples.2.5 discussionPyrosequencing of RdRp gene fragments from coastal samples uncovered much greatergenetic diversity than in previous gene surveys (Culley et al., 2003; Culley and Steward,2007; Culley et al., 2014) and revealed many previously unknown taxonomic groupswithin the Picornavirales. As well, striking dišerences in the taxonomic richness amongsamples implies that these viruses infect a wide variety of eukaryotic plankton, but thatthe mortality imposed on some taxa is highly variable across space and time. Othertaxonomic groups within the Picornavirales were more widespread, suggesting that in-fection of some planktonic taxa is more widespread and persistent. ese results andtheir implications are discussed in detail below.2.5.1 Expanding the known diversity of Picornaviralese high depth of sequencing and limited diversity in each library (Figure 2.3) giveshigh condence that the population structure of RdRp amplicons in each sample hasbeen well characterized (Kemp and Aller, 2004). Although some sequences were closelyrelated to those found in previous studies (Culley et al., 2003; Culley and Steward, 2007)(Figure 2.6), many OTUs formed new clades. Many OTUs were related to Heterosigmaakashiwo RNA virus (HaRNAV) that infects the toxic bloom-forming raphidophyteHet-erosigma akashiwo (Tai et al., 2003). HaRNAV is the type virus of the familyMarnaviri-dae (Lang et al., 2004); it has a genome of about 9.1kb and a high burst size as indicatedby the large crystalline arrays of particles in the cytoplasm of infected cells (Tai et al.,2003). HaRNAV was isolated from coastal waters in British Columbia, Canada (Taiet al., 2003) and can remain infectious formany years in sediments (Lawrence and Suttle,2004). Interestingly, HaRNAV was isolated from the same area as the present study, and47chapter 2appeared ancestral to many of the recovered sequences based on the phylogeny. Forexample, OTU0wasmost abundant (18,034 reads aŸer rarefaction) and clustered closelywith HaRNAV, although the sequence was only 76.4% similar at the amino-acid level.However, other OTUs in the cluster ranged between 55 and 79% similar to HaRNAV,which is low compared to amino-acid similarities of other RNA viruses within a familythat usually have greater than 90% aa similarity (Ng et al., 2012).2.5.2 Distinct communities occurred in dišerent seasons at the same locationWhile only 8.6% of theOTUs from Jericho Pier were shared between dates, both sampleshad similar evenness, although the summer sample had greater richness (Figure 2.5,Table 2.2). e small overlap in OTUs between sampling dates is not surprising giventhe very dišerent conditions between July and October (Figure 2.1), and the dynamicnature of planktonic communities in response to environmental changes. At the samelocation, dsDNA viruses belonging to the Phycodnaviridae, which infect eukaryotic phy-toplankton, varied seasonally based on ngerprint analyses of DNA polymerase genefragments using denaturing gradient gel electrophoresis; however, someOTUs persistedfor extended periods (Short and Suttle, 2003). Similarly, the composition of other aquaticviral communities has been shown to be dynamic although someOTUs persist (Djikenget al., 2009; Rodriguez-Brito et al., 2010), and in some cases have repeatable seasonalpatterns (Chow and Fuhrman, 2012; Clasen et al., 2013; Marston et al., 2013). With onlysingle samples from summer and fall, inferences about dynamics cannot be made fromour data. One of the few taxonomic groups that occurred in both the summer (JP-S)and fall (JP-F) samples from Jericho Pier was related to HaRNAV, (Figure 2.6). erewas greater diversity of OTUs in this clade in JP-F, even though JP-S had higher richnessand higher phylogenetic diversity overall. is is unlike bacterial and phytoplanktoncommunities that tend to be more diverse in winter (Zingone et al., 2009; Ladau et al.,2013). However, the RdRp primers target a specic subset of the viral community thatdoes not režect the overall taxonomic diversity.48chapter 2Based on genome organization and sequence identity, RNA viruses that infect di-atoms have been assigned to the genus Bacillarnavirus, that includes Rhizosolenia setig-eraRNA virus (RsRNAV) (Nagasaki et al., 2004),Chaetoceros tenuissimusMeunier RNAvirus (CtenRNAV) (Shirai et al., 2008) and Chaetoceros socialis f. radians RNA virus (Cs-frRNAV) (Tomaru et al., 2009). In the JP-F sample, the most relatively abundant clustergrouped with RsRNAV that infects the marine diatom Rhizosolenia setigera (Nagasakiet al., 2004).is corresponded with the highest levels of nitrate + nitrite, which is oŸenassociated with high diatom abundances (Zingone et al., 2009); hence, these OTUs arelikely associated with viruses infecting diatoms.2.5.3 Distinct communities occurred at geographically proximate sitesAreas of higher habitat diversity, such as stratied water layers, generally have higherbiological richness (Klopfer andMacArthur, 1960; Chesson, 2000), and this is consistentwith the much higher richness and phylogenetic diversity found at SOG-4, which wasthe most stratied site and included the most abundant OTUs from SOG-1 and SOG-2. Most OTUs from SOG-1 clustered in the Marnaviridae clade, while most SOG-2OTUs clustered in a phylogenetically distant clade. Given that we have used very con-servative clustering, that dsDNA viruses infecting phytoplankton are strain specic andhave phylogenies that are congruent with their hosts (Clasen and Suttle, 2009; Bellecet al., 2014), and that RNA viruses infecting diatoms and the dinožagellate,Heterocapsacircularisquama (Nagasaki et al., 2005) are host-specic, it implies that closely relatedOTUs infect closely related taxa of phytoplankton. Hence, it suggests that the mostabundant viruses at these three locations infect dišerent species.ere are few clear patterns in the spatial distribution of viruses in marine waterswhere geographically distant sites are connected by currents andmixing.e best exam-ples are for cyanophages. For instance, when looking at local variation in cyanophagesisolated at sites in Southern New England, 72% of the viral OTUs were shared betweenat least 2 sites (Marston et al., 2013); however, between Bermuda and Southern New49chapter 2England only 2 OTUs overlapped and they comprised only 0.6% of the isolates. Yet,clear patterns of cyanophage OTU distribution by depth occurred in areas adjacent tothe SOGwhen assessed using community ngerprinting (Frederickson et al., 2003).ebiggest dišerences with depth occurred in stratied water in which some OTUs werepresent at all depths, while others were only present at specic depths, even thoughthe samples were collected only meters apart (Frederickson et al., 2003). ese virusesinfect cyanobacteria, as opposed to the picorna-like viruses, which likely infect protistanplankton. Nonetheless, the factors governing the distribution of cyanobacterial andprotistan hosts are likely similar; hence, dišerent OTUs would be expected to occur inenvironments with dišerent vertical structure (stratication) of the water column.Rank abundance curves showed that SOG-1 and SOG-2 were the least even commu-nities (Figure 2.5). Overall, at most sites four to ve viral OTUs were most abundant(Figure 2.5) similar to other reports for aquatic viral communities in which a few virusesdominate, but most of the diversity comes from rarer viruses (Angly et al., 2006; Suttle,2007). Our targeted approach showed that the picornavirus-like virus communities atSOG and JPwere dominated by only a few genotypes, supporting previousmetagenomicresults showing that the OTU distributions of RNA viruses in SOG and JP were highlyuneven with little overlap between sites (Culley et al., 2006).2.5.4 Each OTU likely represents a single lytic eventGiven that the known hosts of marine Picornavirales isolates targeted by the primerset (Culley and Steward, 2007) are protists, and that protists are the most abundanteukaryotes in the sea, it is likely that the majority of OTUs recovered in this study arefrom viruses that infect these unicellular marine eukaryotes.ese eukaryotic commu-nities are highly dynamic and change throughout the year based on environmental andbiological factors (Larsen et al., 2004). Since viral infection is usually host specic, thediversity in marine viral communities is a režection of the underlying diversity of themarine eukaryotic hosts. Moreover, viral propagation is dependent on host encounter50chapter 2rates and is proportional to host-cell abundance (Murray and Jackson, 1992); hence themost abundant taxa will be most likely to encounter and propagate a viral infection, giv-ing the opportunity for rarer species to increase in abundance and promoting diversity(ingstad, 2000; Winter et al., 2010). Since our study was not over time it is di›cultto evaluate whether these data support the bank model (Breitbart and Rohwer, 2005),however, some taxa were found at one site, but not a similar nearby site, thus these taxacould be present at background levels at some sites and more abundant in others.It is probable that themost abundant OTUs in these data are from recent lysis of hosttaxa. An error rate for replication of RNA viruses of about 1 bp mutation per generation(9000 bp genome x 0.0001 error rate per base pair = 1bp (Holmes, 2009)), and a lower-end burst-size estimate of 1000 particles for marine viruses in the Picornavirales thatinfect protists (Lang et al., 2009), would produce about 1000 dišerent genomes fromeach lysed cell. For the amplied 500bp RdRp gene fragment there is a 0.00056% chanceof an error in 1 generation, assuming thatmutations are distributed evenly in the genome(Sanjuan et al., 2010; Combe and Sanjuán, 2014). Consequently, even with the relativelyhigh error rates of RNA replication, when grouped at 95% similarity at the amino acidlevel, all of the sequences from a lytic event should fall within a single OTU.e half-life for decay of viral infectivity and particles in the surface mixed layer is typically afew hours (Heldal and Bratbak, 1991; Suttle and Chen, 1992; Noble and Fuhrman, 1997;Bettarel et al., 2009); thus the recovered viral OTUs were likely from recent lytic events.Furthermore, considering the specicity of viruses infecting protists (Short, 2012), eachOTU probably stems from viruses infecting a single host taxon.us, these data implythat infection of marine protists by viruses in the Picornavirales is not only pervasive,but likely involves a wide diversity of host taxa; hence, these viruses are likely importantstructuring elements for phytoplankton communities that inžuence nutrient cycling andenergy žow.51chapter 22.5.5 Amplicon deep sequencing as an approach for estimating viral diversityAmplicon deep sequencing is a sensitive and high-resolution approach for examiningmicrobial community dynamics over time and space (Gobet et al., 2012; Caporaso et al.,2011; Gibbons et al., 2013). Careful quality trimming of sequences and removal of sin-gletons is essential for reliable results (Zhou et al., 2011) since errors in sequences willinžate estimates of diversity. With careful data processing and analysis, amplicon deepsequencing is as accurate for assessing community composition and diversity as cloningand Sanger sequencing (Amend et al., 2010), but with much greater depth of coverageof the community.ere are potential biases associated with reverse transcription with random hexam-ers (which can decrease yield and could inžate diversity) (Zhang and Byrne, 1999), tem-plate amplication by PCR (Lee et al., 2012) and with using highly degenerate primersthat target a specic part of the community containing many dišerent templates (Cul-ley and Steward, 2007). A danger of the high cycle number can be diversity overesti-mates which can come from the increasing number of chimeric sequences producedwith greater cycle number (Qiu et al., 2001).e sequences were processed with cautionconsidering the high number of PCR cycles employed in this study. Chimera checkingdenovo was used to look for chimeric sequences originating from two higher abundancereads, and reference-based chimera checking was used a database of RdRps from iso-lated viruses to correct for this potential error. In addition, a conservative cut-oš wasused of only OTUs comprising more than 5 reads that aligned to the conserved domainalignment.Although read abundance of OTUs can be considered semi-quantitative and goodfor comparisons of richness and diversity among samples (but not for absolute counts ofgenes)(Amend et al., 2010; Pinto and Raskin, 2012; Ibarbalz et al., 2014). Moreover, byusing control sequences obtained by cloning and Sanger sequencing alongside pyrose-quenced libraries containing the same sequence (Appendix A) we veried that amplicondeep sequencing and our sequence processingmethodology recovered accurate environ-52chapter 2mental viral sequences and non-inžated estimates of richness like in studies for bacterialamplicons (Sogin et al., 2006; Huse et al., 2008; Kirchman et al., 2010; Caporaso et al.,2011) and clinical viral studies (Watson et al., 2013; Romano et al., 2013).2.6 conclusionAmplicon deep sequencing of RdRp gene fragments using 454 pyrosequencing revealedthe richness and population structure of marine Picornavirales in ve coastal samples.e known diversity of viruses in this group was greatly increased with 145 OTUs thatdišered by at least 5% at the amino-acid level. ere were between 24 and 71 OTUs ineach sample, with distinct patterns of OTU distribution, richness and diversity amongsamples. ere was little overlap between viral OTUs collected at the same site in sum-mer and fall, and among samples collected 20 km apart on the same day.e high tem-poral and spatial diversity in RdRp sequences is consistent with viral communities thatturnover rapidly, and episodic infection of a wide diversity of protistan hosts. e lowoverlap inOTUs and phylogenetic diversity among samples implies a dynamic landscapeof viral infection and supports the idea that marine picorna-like viruses are importantpathogens of marine protists that have an important role in structuring marine plank-tonic communities, and in nutrient cycling and energy transfer among trophic levels.Ultimately, further study is needed to disentangle the temporal and spatial drivers ofthese communities.53chapter 3Marine virus and host community structure exhibitstemporal phylogenetic dynamics3.1 summaryMarine microbes and their viruses are essential parts of the marine ecosystem that formthe base of the foodweb, and drive biogeochemical cycles. Studies have shown that ma-rine viral communities display repeatable changes in abundance and community com-position with time; however, whether these changes režect shiŸs in dominance withinevolutionarily related groups of viruses and their hosts is unexplored. To examine thesedynamics, changes in the composition and phylogenetic makeup of two ecologicallyimportant groups of viruses, and their potential hosts, were followed at a coastal sitenear Vancouver, Canada, every two weeks for 13 months. Changes in the taxonomiccomposition within DNA bacteriophages belonging to the T4-like myoviruses and ma-rine picorna-like RNA viruses infecting eukaryotic phytoplankton, as well as bacteriaand eukaryotes, were followed using amplicon sequencing of gene fragments encodingthe major capsid protein (gp23), the RNA-dependent RNA polymerase (RdRp) and the16S and 18S ribosomal RNA genes, respectively. e results showed that for the viralgroups the dominant groups of phylogenetically related viruses shiŸed over time, andthat there were many transient taxa and few persistent taxa. Yet, dišerent communitystructures were observed for dišerent marker genes. Additionally, with strong laggedcorrelations between viral richness and community similarity of putative hosts, the re-sults imply that viruses inžuence the composition of the host communities, and that54chapter 3their community structure is dependent on lifestyle, cementing their role as importantstructuring elements in marine planktonic communities.3.2 introductionUnderstanding diversity, its maintenance and drivers is a continued theme in ecology.is is very evident formicrobial systems, for which there has been extensive explorationand discussion on the mechanisms responsible for the observed high diversity. Manystudies on microbial diversity and dynamics come from the marine milieu, where it hasbeen argued that community composition is driven by environmental factors (DuRandet al., 2001; Morris et al., 2005; Fuhrman et al., 2006; Gilbert et al., 2009). Against thisbackdrop are viruses, which are obligate and ubiquitous, and the most abundant anddiverse biological entities in the world’s oceans (Suttle, 2005).is diversity arises since viruses have many dišerent lifestyles (Paul, 2008), andmorphologies (Brum et al., 2013). Diversity is also generated from an assortment ofinfection strategies as some viruses infect specic strains or species of hosts whereasothers have broad host-ranges (Breitbart, 2012). As well, some groups of viruses showparticularly high genetic diversity because of their low delity of replication (Lang et al.,2009), while others have high rates of horizontal gene transfer (e.g. Moreau et al., 2010).e role of viruses as obligate pathogens with high host specicity implies that theyare important drivers of host composition and diversity (Rodriguez-Valera et al., 2009);yet, our understanding of their roles as drivers of marine microbial diversity remainsrelatively unexplored.Marine viruses have repeatable seasonal dynamics as revealed by measures of abun-dance, infectious units, and taxonomic composition. Seasonal studies in coastal watershave reported that viral abundances are higher in summer than in winter (Bergh et al.,1989; Jiang and Paul, 1994), while multi-year time time series data show that viral pro-duction and viral abundances are highest in early spring and summer (Winget et al.,55chapter 32011; Parsons et al., 2012) although viral lysis is highest in winter (Winget et al., 2011).Moreover, viral dynamics can be associated with putative hosts (Parsons et al., 2012)and specic subsets of the coastal viral community can show seasonal community com-position dynamics (Chow and Fuhrman, 2012; Pagarete et al., 2013). As well, virusesinfecting cyanobacteria are also temporally dynamic (Waterbury and Valois, 1993; Suttleand Chan, 1994), with communities from the same season resembling each other morethan communities sampled in the same year (Marston et al., 2013) and winter communi-ties being more stable than in the summer and spring (Clasen et al., 2013).Viruses ašect community composition in laboratory studies by reducing the abun-dance of the dominant host, allowing others to grow up (e.g. Middelboe, 2000; Mid-delboe et al., 2001; Bouvier and del Giorgio, 2007; Marston et al., 2012); thus, virusespromote diversity at the strain level and can be responsible for large shiŸs in bacterialpopulations (Hewson et al., 2003; Schwalbach et al., 2004; Rodriguez-Valera et al., 2009).ese dynamics have been termed “Killing the Winner” (KtW), a theory in which themost actively growing hosts are killed by viruses only to be replaced by another strainor species (ingstad, 2000;ingstad et al., 2015). ere is evidence that these KtWdynamics occur in the eld as illustrated by a study in a solar saltern where coarselydened bacterial and viral taxa were relatively constant over time, but showed KtWdynamics at a ner scale (Rodriguez-Brito et al., 2010). However, few environmentalstudies have shown evidence for these dynamics since few have compared hosts andviruses (Winter et al., 2010).Examining the temporal dynamics of marine viruses and their hosts has yieldedinsights about their ecology and evolution, yet little attention has been paid to the phylo-genetic relationships within these communities and how they are shaped. An exceptionis a study by Goldsmith et al. (2015), near Bermuda, where the phylogenetic makeupof related groups of viruses over time and depth was found to be highly uneven andvariable. ere were dišerences between fall and winter attributable to stratication,with much of the variability due to one phylogenetic group of cyanophages (Goldsmith56chapter 3et al., 2015). Clasen et al. (2013) also found that groups of cyanophages belonging todišerent phylogenetic clades shiŸed in their relative dominance over time. Knowingmore about the phylogenetic diversity of the viral communities will allow us to betterinterpret these temporal dynamics.Phylogenetic relatedness can be correlated to ecological relatedness in plants andanimals (Harvey and Purvis, 1991; Srivastava et al., 2012) and microbes have shownphylogenetic patterning (Horner-Devine and Bohannan, 2006; Lennon et al., 2012), yetlittle is known about these patterns in viral communities. To examine these patterns overtime, the following was hypothesized. First, since viruses can be following their hosts(Chow et al., 2014) or being driven by Killing theWinner dynamics (ingstad, 2000), itwas hypothesized that phylogenetic patterns would be detected in the viral communities,as was found in the putative host communities. Second, the structure and compositionof viral communities follows a “seed bank” distribution where there are many more rareviral operational taxonomic units (OTUs) than abundant ones (Breitbart and Rohwer,2005; Goldsmith et al., 2015), therefore, a shuœing in rank of related viral OTUs wouldbe predicted over time, but with a few OTUs dominating the community.To test these hypotheses, the temporal dynamics of the phylogenetic make-up oftwo ecologically important groups of marine viruses and their potential hosts were fol-lowed in samples taken every two weeks for thirteen months, using amplied markergenes and high-throughput sequencing. With these data, the hypotheses were testedby looking at the community similarities, comparing the phylogenetic diversity overtime, and observing the relative abundance of phylogenetically-related groups of OTUsover time in viral and putative host communities.e rst group of viruses was the T4-like myoviruses, which are DNA viruses that infect bacteria, including cyanobacteria.A structural gene, gp23, which encodes the capsid was used as an amplication target(Filée et al., 2005). e second was a group of eukaryote-infecting RNA viruses: themarine picorna-like viruses, which were targeted by amplifying the RNA dependentRNA polymerase (RdRp) (Culley and Steward, 2007).ese RNA viruses infect ecolog-57chapter 3ically important phytoplankton such as diatoms belonging to the genera Rhizoseleniasp., Chaetoceros sp., and the toxic bloom-forming raphidophyte Heterosigma akashiwo(Tomaru, 2015).e dynamics of putative hosts were examined by sequencing ampliedmarker genes for eukaryotes (18S rRNA gene) and bacteria (16S rRNA gene). is con-tribution opens up new avenues of understanding by showing that temporal changes inthe phylogenetic make-up of viruses infecting bacteria and eukaryotic algae are relatedto seasonal žuctuations in the communities of potential hosts.3.3 materials and methods3.3.1 Sample collectionSeawater samples were collected from Jericho Pier (49° 16’36.73N, 123° 12’05.41W) inBritish Columbia, Canada. Jericho Pier (JP) is adjacent to the shoreline, in a well-mixedlocation with mixed semi-diurnal tides. In order to get a representative sample of waterand enoughmaterial for viral extraction 60Lofwaterwas pumped from the 1-mdepth ev-ery two weeks at the daytime high tide between June 2010 and July 2011 (33 samples). Anadditional set of seven samples was collected every two days from 29 January to 10 Febru-ary 2011 for more high-resolution analysis of dynamics within these two weeks. Salinityand temperature were measured using a YSI probe (Yellow Springs, Ohio, USA). For allsamples, the water was pre-ltered through a 65µmNitex mesh and ltered sequentiallythrough 142-mm diameter, 1.2-µm nominal pore-size glass-ber (GC50 Advantec MFS,Dublin, CA., USA) and 0.22 µmpore-size polyvinyldine (Millipore, Bedford,MA,USA)lters.e ltrate, containing the viral size fraction, was concentrated to ~500mL (viralconcentrate) using tangential žow ultraltration with a 30kDa MW prep-scale SpiralWound TFF-6 cartridge (Millipore) (Suttle et al., 1991).58chapter 33.3.2 NutrientsPhosphate, silicate and nitrate+nitrite concentrations were determined in duplicate 15-mL seawater samples ltered through0.45 µmpore-sizeHAlters (Millipore) and storedat -20°Cuntil air-segmented continuous-žowanalysis on aAutoAnalyzer 3 (Bran+Luebbe,Norderstedt, Germany). Chlorophyll a (Chl a) was determined in triplicate by ltering100 mL of seawater onto 0.45 µm pore-size HA lters (Millipore), and storing the ltersin the dark at -20°C until acetone extraction and then analysed žuorometrically (Parsonset al., 1984).3.3.3 Enumeration of bacteria and virusesSamples for viral and bacterial abundances were taken at each sampling point by x-ing duplicate cryovials containing 980µL of sample with nal concentration of 0.5%glutaraldehyde (EM-grade), freezing in liquid nitrogen and storing at -80°C until pro-cessing. Flow cytometry samples were processed as in Brussaard (2004b). Briežy, vi-ral samples were diluted 1:10 to 1:10 000 in sterile 0.1 µm ltered 1X TE, stained withSYBR Green I (Invitrogen, Waltham, MA, USA) at a nal concentration of 0.5 x 10-4of commercial stock, heated for 10 minutes at 80° C and then cooled in the dark for 5minutes before processing. Bacterial samples were diluted up to 1:1000 in sterile 0.1 µmltered 1xTE, stained with SYBR Green I (Invitrogen) at a nal concentration of 0.5 x10-4 of commercial stock, and incubated in the dark for 15 minutes before processing.All samples were processed on a FACScalibur (Becton-Dickinson, Franklin Lakes, NewJersey, USA) with viral and bacterial samples run for 1 min at a medium or high žowrate, respectively. Event rates were kept between 100 to 1000 events per second andgreen žuorescence and side scatter detectors were used. Data were processed and gatedusing Cell-Quest soŸware (Becton-Dickinson).59chapter 33.3.4 Extraction of viral nucleic acidse viral concentrate was ltered twice through 0.22 µm pore-size Durapore PVDFlters (Millipore) in a sterile Sterivex lter unit (Millipore).e ltrate, containing viral-sized particles, was pelleted by ultracentrifugation (Beckman-Coulter, Brea, California,USA) in a SW40 rotor at 108 000 g for 5 h at 12°C.e pellet was resuspended overnightin 100 µL of supernatant at 4°C. To digest free DNA, the pellets were incubated with 1UµL-1DNAsewith a nal concentration 5mMMgCl2 for 3 h at room temperature. Nucleicacids were extracted using a Qiamp Viral Minelute spin kit (Qiagen, Hilden, Germany)according to the manufacturer’s directions.3.3.5 PCR amplication of T4-like myoviral marker geneTo target the marine T4-like myoviral capsid protein gene (gp23), PCRs were set up asin Filée et al. (2005). Briežy, each reaction mixture (nal volume, 50 µL) consisted of 2µL template DNA, 1x (nal concentration) PCR bušer (Invitrogen, Carlsbad, California,USA), 1.5 mMMgCl2, 0.2 mM of each deoxynucleoside triphosphate (Bioline, London,UK), 40 pmol of MZIA1bis and 40pmol of MZIA6, and 1 U Platinum Taq DNA poly-merase (Invitrogen) and program conditions as in Table 3.1.Table 3.1: PCR parameters used in this study. (continued below)Markergene TargetPrimernames PCR initialPCRdenaturationAnnealingtemperaturegp23 T4-likemyovirusMZIA1bisandMZIA694°C for 90 s 94°C for 45 s 50°CRdRp Marinepicorna-likevirusesMPL-2FandMPL-2R94°C for 75 s 94°C for 45 s 43°C60chapter 3Table 3.1: continuedMarkergene TargetPrimernames PCR initialPCRdenaturationAnnealingtemperature18S rRNAgeneEukaryotes Euk1209fandUni1392r94°C for 75 s 94°C for 1min65°Ctouchdown for10 cyclesfollowed by55°C16S rRNAgeneBacteria 341F and907R94°C for 75 s 94°C for 1min64°C, 12cyclesfollowed by54°CExtension Cycles Final extension Reference72°C for 45 s 35 5 min at 72°C Filée et al. (2005)72°C for 60 s 40 9 min at 72°C Culley and Steward (2007)72°C for 60 s 10 + 20 9 min at 72°C Diez et al. (2001)72°C for 60 s 12+25 10 min at 72°C Baker et al. (2003) andMuyzer et al. (1995)3.3.6 PCR amplication of picorna-like virus marker geneHalf of each viral extract was used to synthesize cDNA. To remove DNA, the extractedviral pellets were digested with DNase 1 (amplication grade) (Invitrogen).e reactionwas terminated by adding 2.5 mM EDTA (nal concentration) and incubating for 10min at 65°C. Complementary DNA (cDNA) was generated using Superscript III reversetranscriptase (Invitrogen) with random hexamers (50 ng µL-1) as per the manufacturer.61chapter 3PCR was performed using primer set MPL-2 to target the RdRp of marine picorna-like viruses (Culley and Steward, 2007). Each reaction mixture (nal volume, 50 µL)consisted of 50 ng of cDNA, 1x (nal concentration) PCR bušer (Invitrogen), 2 mMMgCl2, 0.2 mM of each deoxynucleoside triphosphate (Bioline, London, UK), 1 µMof each primer, and 1 U Platinum Taq DNA polymerase. e reaction was run in aPCR Express thermocycler (Hybaid, Ashford, UK) with program conditions as in Table3.1. Products were run on a 0.5X TBE 1% low melt gel, excised and extracted usingZymoclean Gel DNA Recovery Kit (Zymo) as per the manufacturer and a nal elutionstep of 2x10 µL EB bušer (Qiagen).3.3.7 Filtration and extraction of marine bacteria and eukaryotesOne liter of seawater was taken from the sixty liters and ltered through a 0.22µm pore-size Durapore PVDF 47 mm lter (Millipore) in a sterile Sterivex lter unit (Millipore).e lter was either stored at -20°C until extraction or immediately extracted as follows.Filter extraction was as in Short and Suttle (2003). Briežy, lters were aseptically cut andincubated with lysozyme (Sigma-Aldrich, St. Louis, MO, USA) at a nal concentrationof 1mg mL-1 for 2 h at 37°C. Sodium dodecyl sulfate was added at a nal concentrationof 0.1% (w/v) and each lter was put through three freeze-thaw cycles. Proteinase K(Qiagen) was then added to a nal concentration of 100 µg mL-1 and incubated for 1 h at55°C. DNA was sequentially extracted using equal volumes of phenol:chloroform:IAA(25:24:1), and chloroform:IAA (24:1). DNA was precipitated by adding NaCl to a nalconcentration of 0.3M and by adding 2X the extract volume of ethanol. Samples wereincubated at -20°C for at least 1 h and then centrifuged for 1 h at 20 000 g at 4°C. Extractswere washed with 70% ethanol and were resuspended in 50 µL EB bušer (Qiagen).3.3.8 PCR amplication of bacterial and eukaryotic ribosomal sequencesPCR targeting eukaryotes used primers Euk1209f and Uni1392r as in Diez et al. (2001).ese primers target positions 1423 to 1641 and includes the variable region V8. Each re-62chapter 3action mixture (nal volume, 50 µL) consisted of 2 µL template, 1x (nal concentration)PCR bušer (Invitrogen), 1.5 mMMgCl2, 0.2 mM of each deoxynucleoside triphosphate(Bioline, London, UK), 0.3 µMof each primer, and 2.5 UPlatinumTaqDNApolymerase.e reaction was run in a PCR Express thermocycler (Hybaid, Ashford, UK) with pro-gram conditions as in Table 3.1.PCR targeting bacteria used primers 341F (Baker et al., 2003) and 907R (Muyzeret al., 1995).ese primers target the v3 to v5 regions. PCRs were run with the followingconditions: each reaction mixture (nal volume, 50 µL) consisted of 2 µL template, 1x(nal concentration) PCR bušer (Invitrogen), 1.5 mMMgCl2, 0.2 mM of each deoxynu-cleoside triphosphate (Bioline, London, UK), 0.4 µM of each primer, and 1 U PlatinumTaq DNA polymerase. e reaction was run in a PCR Express thermocycler (Hybaid,Ashford, UK) with program conditions as in Table 3.1.3.3.9 Sequencing library preparationConstruction — PCR products not requiring gel excision were puried aŸer PCRusing AMPure XP beads (Beckman Coulter) at a ratio of 1.2:1 beads:product. Cleanedproducts were resuspended in 30 µL EB bušer (Qiagen). All products were quantiedusing the Picogreen dsDNA (Invitrogen) assay using Lambda DNA (Invitrogen) as astandard. Sample concentrations were read using iQ5 (Bio-Rad, Hercules, CA, USA)and CFX96 Touch systems (Bio-Rad). Pooled libraries were constructed using one ofeach of the amplicons at a concentration so that their molarity would be similar andthe total product of the pool to be ~700-900 ng. Pooled amplicons were concentratedusing AMPure XP beads (Beckman Coulter) at a ratio of 1.2:1 beads:product. NxSeqDNA sample prep kit 2 (Lucigen, Middleton, WI, USA) was used as per manufacturer’sdirections with either NEXTFlex 48 barcodes (BioO, Austin, USA), NEXTžex 96 HTbarcodes (BioO), or TruSeq adapters (IDT, Coralville, Iowa). Libraries were cleaned upusing AMPure XP beads (Beckman Coulter) at a ratio of 0.9:1 beads:library.63chapter 3Quantification and quality control of libraries — Libraries were checked forsmall fragments (primer dimers and/or adapter dimers) using a 2100 Bionanalyzer (Ag-ilent, Santa Clara, CA, USA) with the High Sensitivity DNA kit (Agilent). e concen-tration of libraries was quantied using Picogreen dsDNA assay as above.e librarieswere quantied and checked for ampliable adapters using the Library QuanticationDNA standards 1-6 (Kappa Biosystems, Wilmington, USA) with the SsoFast EvaGreenqPCR supermix (Bio-Rad) using 10 µL EvaGreenmaster mix, 3 µL of 0.5 µMF primer, 3µL of 0.5 µmRprimer and 4 µL of 1:1000, 1:5000 and 1:10000 dilutions of the libraries intriplicate on iQ5 (Bio-Rad) andCFX96Touch qPCRmachines. Cycling parameters wereas follows: 95°C for 30s, 35 cycles of 95°C for 5s, 60°C for 30s, and the melt curve gener-ation from 65°C to 95°C in 0.5°C steps (10s/step). Quantication from both Picogreenand qPCR assays were used to determine nal pooling of all libraries before sequenc-ing. Libraries were sequenced using 2x250bp PE Miseq (Illumina, San Diego, USA)sequencing at Génome Québec Innovation Centre at the McGill University (Montreal,QC, Canada), and 2x300bp PE Miseq (Illumina) sequencing at UBC PharmaceuticalSciences Sequencing Centre (Vancouver, BC, Canada) and at UCLA’s Genoseq (LosAngeles, CA, USA).3.3.10 Initial sequence processingLibraries were either split by the sequencing centre using CASAVA (Illumina) or split bythe user using the Miseq Reporter soŸware (Illumina). Sequence quality was initiallyexamined using FastQC (Andrews, 2015). Contaminating sequencing adapters wereremoved using Trimmomatic version 0.32 (Bolger et al., 2014) and the quality of thesequencing library further examined using fastx_quality (Gordon, 2014). Libraries werefurther split into individual amplicons (i.e. 18S, 16S, gp23 and MPL) and then, if the ex-pected overlap of the paired-end reads was 40bp or more, the paired reads were mergedusing PEAR (Zhang et al., 2014). Sequences were then quality trimmed using Trimmo-matic with the default quality settings. Sequences were aligned to known sequences64chapter 3(Silva 119 database (Quast et al., 2013) for 16S and 18S rRNA genes) using align.seqs inmothur 1.33.3 (Schloss et al., 2009) and those not aligned were removed. Viral sequenceswere queried using BLAST against databases containing the genemarkers of interest andsequences with an e-value below 10-3 were kept.3.3.11 Chimera checking, OTU picking and read normalizatione 16S and 18S rRNA gene sequences were checked for chimeras using USEARCHversion 8.0.1517 reference (Edgar, 2010) with the Gold reference database. Unique, non-chimeric sequenceswere clustered at 97% similarity. Taxonomy for the 16S and 18S rRNAgene sequences was assigned usingmothur (Wang-type algorithm) and the taxonomy inSilva 119 (Quast et al., 2013). For the viral targets sequences were chimera-checked usingUSEARCH denovo and reference (Edgar, 2010). Viral sequences were then translatedusing FragGeneScan 1.20 (Rho et al., 2010). Viral reads were clustered using USEARCH(Edgar, 2010) at 95% similarity for MPL, and 95% similarity for T4-like myoviruses. Op-erational taxonomic unit (OTU) tables for all targets were constructed using USEARCH(Edgar, 2010). Rarefaction curves were generated using vegan (Oksanen et al., 2015).Sequences were normalized for this project by date and by target using vegan (Oksanenet al., 2015).3.3.12 Data analysis and multivariate statisticsEnvironmental data — Environmental parameters were mean imputed to ll indata missing because of instrument malfunction or unavailability. Day length data wereretrieved using R package geosphere (Hijmans, 2015).Community similarityandMantel test — Spearman rank correlations using rcorras part of Hmisc (Harrell, 2014). Adonis was used from vegan (Oksanen et al., 2015) totest whether community matrices showed seasonal dišerences. Bray-Curtis distancematrices were constructed from the normalized OTU abundance tables. Mantel tests65chapter 3were performed by comparing the community distance matrices to each other and todistance matrices of environmental parameters using vegan.3.3.13 PhylogenyNCBI CDD domain alignments for RdRp and for gp23 were retrieved and used as hid-den markov models via HMMER (Johnson et al., 2010) to align translated OTUs withClustal Omega (Sievers et al., 2014). Environmental sequences for both the gp23 (Filéeet al., 2005; EF617478 Sandaa and Kristiansen, 2007; Jia et al., 2007; López-Bueno et al.,2009; ACU57502-AC57509 Mabizela and Litthauer, 2009; Comeau et al., 2010; Chowand Fuhrman, 2012; Bellas and Anesio, 2013; Butina et al., 2013) and the RdRP (Culleyet al., 2003; Culley and Steward, 2007) were retrieved from Genbank to give context tothe OTUs.Alignments were checked and manually curated with aliview (Larsson, 2014). Au-tomated trimming of the alignment was done using Trimal (Capella-Gutierrez et al.,2009). Model VT was chosen for the RdRP gene and model JTT for the gp23 gene usingProttest (Darriba et al., 2011). Initial phylogenetic trees were built with Fast Tree (Priceet al., 2010) and examined in FigTree (Rambaut, 2014). Final maximum likelihood treeswere generated using RAxML (Stamatakis, 2014) with 1000 bootstraps, with VTwith thePROTGAMMAmodel for the RdRp gene tree and JTT with the PROTGAMMAmodelfor the gp23 gene tree on the CIPRES webserver (Miller et al., 2010). Faith’s phylogeneticdiversity (Faith, 1992) was calculated as implemented in picante (Kembel et al., 2010).e package ggtree (Yu et al., 2016) was used for visualizing and annotating trees andggplot2 (Wickham, 2009) was used for all other plots made in R (R Core Team, 2014).All scripts used for processing the data are available on github (Gustavsen, 2016).66chapter 3Figure 3.1: Chlorophyll a concentration over time at Jericho Pier, Vancouver, BritishColumbia. Inset: Chlorophyll a concentration from 0-50 ug/L.67chapter 3Figure 3.2: Environmental parameters during 1-year time series at Jericho Pier. Viraland bacterial abundance were measured by žow cytometry. Grey vertical lines indicateseason boundary and green vertical line indicates time of spring bloom. Error bars arestandard error based on duplicates.68chapter 33.4 results3.4.1 Variability of environmental characteristicsChlorophyll a (chl a) concentrations varied over time with a maximum observed con-centration during a eukaryotic phytoplankton bloom (46.5 µg L-1 in June 2011)(Figure3.1). e second highest chl a occurred during the annual spring bloom in late April2011 (5.88 µg L-1) which is mainly composed of diatoms belonging toallassiosira sp.(Harrison et al., 1983; Allen and Wolfe, 2013). e minimum chl a value of 0.05 µg L-1occurred in May and the chlorophyll levels remained below 1 µg L-1 from September toMarch.Nutrient concentrations were also highly dynamic ranging between 6.1 µM to 67.3µM for silicate, from below 0.1µM to 2.3 µM for phosphate and from below 0.1µM to27.7 µM for nitrate+nitrite (Figure 3.2). Overall, nutrient concentrations were high andstable over winter, dipped in late April and then were followed by a large increase insilicate commencing in May.e viral abundance ranged from 5.41 × 106 to 4.695 × 107 particles mL-1 while thebacterial abundance was the expected one order of magnitude lower of 6.59 × 105 to4.43 × 106 cells mL-1 (Figure 3.2).3.4.2 Richness and shared microbial and viral OTUs over timeFor each amplicon target representing T4-like myoviruses (gp23), picorna-like viruses(RdRp) bacteria (16S rRNA gene) and eukaryotes (18S rRNA gene), the sequences weretranslated into amino acids (except 16S and 18S) and normalized to the library withfewest reads. is resulted in 1737 OTUs from T4-like myoviruses at 95% sequencesimilarity, with a maximum of 495 and a minimum of 149 per timepoint. On average6% of the T4-like myoviruses OTUs were shared among times. For the marine picorna-like viruses there were 574 OTUs at 95% sequence similarity, with between 60 to 149OTUs per timepoint. On average 6% of these OTUs were shared. ere were 802 bac-69chapter 3terial OTUs (97% sequence similarity) with an average of 10% shared over time. ehighest number of OTUs seen per time point was 270 and the minimum was 82. In theeukaryotic community a total of 1117 OTUs (97% sequence similarity) were found with6% shared on average, a maximum of 297 and a minimum of 62.Rarefaction curves for individual samples did not žatten (“saturate”) indicating thatnot all possible OTUs were sequenced in these samples, but when considered togetherthe curves saturated indicating that even if all the diversity was not captured in onesample, the overall community of OTUs was captured (Figure 3.3).3.4.3 Community similarity, phylogenetic diversity, and richness over timeSpecies richness (SR) and phylogenetic diversity (PD) were stable for the bacteria (Fig-ure 3.4C) and eukaryotes (Figure 3.4D) from the fall to winter months, although therewere marked changes in the similarity and richness of the bacterial community duringFebruary (Figure 3.4C) even though environmental conditions were relatively stable(Figure 3.2). AŸer this change, bacterial diversity began decreasing through to July. Foreukaryotes, the SR and PD decreased aŸer December until February and then climbedagain until mid-March and then decreased until mid-May. e marine picorna-likeviruses generally had congruent patterns in PD and SR, but during the spring bloom(alassiosira sp.) the SR and PD diverged and the PD was among the highest values ob-served. February was the time of highest SR and PD for the T4-like myoviruses, but wasfollowed by the lowest values. In contrast to the bacteria, in which richness decreasedaŸer the spring bloom, the richness of the T4-like myoviruses increased. Likewise, thespring and summer T4-like myovirus similarity lagged behind that of the bacteria.3.4.4 Dynamics of phylogenetically-related viral OTUsTo understand their dynamics viral OTUs were placed into a phylogenetic context (Fig-ure 3.5). Well-dened andwell-supported phylogenetic groups (A-H) ofmarine picorna-like viruses (Figure 3.5A andB) showed strong temporal dynamics and dišered by season70chapter 3Figure 3.3: Rarefaction curves of samples from Jericho Pier time series. A) Rarefactioncurve 18S rRNA gene, B) Rarefaction curve 16S rRNA gene, C) Rarefaction curve T4-likemyovirus(gp23), D) Rarefaction curve Marine picorna-like(RdRp).71chapter 3Figure 3.4: Species richness, phylogenetic diversity, and community similarity overtime. A) T4-like myoviruses (gp23), B) Marine picornalike-viruses (RdRp), C) bacteria(16S rRNA gene), and D) eukaryotes (18S rRNA gene). Faith’s phylogenetic diversity(Faith, 1992)was calculated as implemented in picante (Kembel et al., 2010). Communitysimilarity is Bray-Curtis calculated on normalized data between sequential times. Greyvertical lines indicate season boundary and green vertical line indicates time of springbloom.72chapter 3Figure 3.5: Maximum likelihood RAxML phylogenetic trees and barplots of closely-related phylogenetic groups of OTUs. A) Tree of marine picorna-like virus RdRpsequences including reference sequences and OTUs generated in this study. Outgroupis virus Equine rhinitis B virus (Picornaviridae). B) Barplot of the relative abundances ofmarine picorna-like virus phylogenetic groups over time. C) Tree of T4-like myovirusmajor capsid protein sequences including reference sequences and OTUs generated inthis study. Outgroup is Enterobacteria phage T4. D) Barplot of the relative abundancesof T4-like myovirus phylogenetic groups from over time. Grey vertical lines indicateseason boundary and green vertical line indicates time of spring bloom.73chapter 3Figure 3.6: Maximum likelihood phylogenetic tree (RAxML) of marine picorna-likeviruses including reference sequences and OTUs. Outgroup is virus Equine rhinitis Bvirus (Picornaviridae). OTUs at 95% similarity at the amino-acid level. Detailed sub-trees with tip labels are availabe in Appendix B.74chapter 3Figure 3.7: Maximum likelihood phylogenetic tree (RAxML) of T4-like myovirusesincluding reference sequences and OTUs. Outgroup is Enterobacteria phage T4. OTUsat 95% similarity at the amino acid level. Detailed sub-trees with tip labels are availabein Appendix B.75chapter 3Figure 3.8: Barplot of top 20 most relatively abundant marine picorna-like virus OTUs.Relative abundance is the proportion of the community aŸer overall normalization bysite. Grey vertical lines indicate boundaries between seasons and the green vertical lineindicates the time of the spring bloom.76chapter 3Figure 3.9: Barplot of the top 20 most relatively abundant T4-like myovirus OTUs.Relative abundance is the proportion of community aŸer overall normalization by site.Grey vertical lines indicate boundaries between seasons and the green vertical lineindicates the time of the spring bloom.77chapter 3(Adonis R= 0.506, p-value 0.001). Group H, which includes viruses that infect the di-atoms Chaetoceros sp., and Rhizoselenia sp., were constant members of the communitiesalthough their relative abundance was highest in late November. Group A was alwayspresent and includes many environmental sequences, as well as a virus that infects theraphidophyte,Heterosigma akashiwo; the groupwasmost abundant betweenAugust andSeptember. e months from October to February were dominated by OTUs in groupE, with a smaller contribution by group H.e structure of the MPL OTU phylogeneticgroups closely mirrored the structure of the top 20 OTUs found over time (Figure 3.8),demonstrating that this community contained few dominant OTUs.e T4-like myoviruses OTUs were also placed in a phylogenetic context and cate-gorized into groups of related OTUs (Figure 3.5 C). In the fall, group I dominated thecommunity, followed by group G; both groups include viral isolates infecting cyanobac-teria. In January almost half of the relative abundance of the T4-like myoviruses was rep-resented by group B which contains no known isolates. Unlike the marine picorna-likeviral community, the T4-likemyoviral community had very dišerent patterns among thetop 20OTUs and the phylogenetic groups over time (Figure 3.9). When there was a largeincrease in nutrients in late September (Figure 3.2), there were shiŸs in the dominantgroups in the T4-like myoviral community. e community returned to its previousstate by the next sampling time. e communities showed small dišerences by season(Adonis R= 0.231, p-value 0.001)3.4.5 Taxonomic richness of bacteria over timeAt the phylum level the bacterial communities were relatively stable over time (Figure3.10).e communities were dominated by Proteobacteria and at times had large propor-tions of Bacteroidetes.e phylum Bacteroidetes was mostly dominated by members ofthe Flavobacteria with Cytophaga and Bacteroidia in the fall, and Sphingobacteriia inthe winter. e phylum Actinobacteria was present and included stable populations ofAcidomicrobia and Actinobacteria, which were mostly absent in March. AŸer June the78chapter 3Figure 3.10: Bacterial OTUs by phylum over time. Classications were done using theWang algorithm as implemented in mothur (Schloss et al., 2009) and using the Silva 119database (Quast et al., 2013). Grey vertical lines indicate boundaries between seasonsand the green vertical line indicates the time of the spring bloom.79chapter 3Actinobacteriamake up a larger proportion of the community.e bacterial communityshowed small dišerences by season (Adonis R: 0.361, p-value 0.001).3.4.6 Taxonomic richness of eukaryotes over timeFigure 3.11: Eukaryotic OTUs by phylum over time. Classications were done using theWang algorithm as implemented in mothur (Schloss et al., 2009) and using the Silva 119database (Quast et al., 2013). Grey vertical lines indicate boundaries between seasonsand the green vertical line indicates the time of the spring bloom.A large portion of the eukaryotic community was made up of Opistokonts, and alsomembers of the SAR supergroup (Figure 3.11). Haptophytes were also important mem-bers of the community from December to January and March to April. Cryptophyteswere also relatively abundant in late December to early January andMarch. At the OTUlevel the community dišered by season (Adonis R=0.23, p-value 0.001).80chapter 33.4.7 Heatmaps of persistent vs. ephemeral OTUsFigure 3.12: Plot of relative abundance of T4-like myoviral OTUs (95% amino acidsimilarity) that were present in over 90% of samples (persistent) or in less than 20% ofsamples (ephemeral). OTUs are ordered by phylogenetic tree. Stars (*) indicate missingsamples. A heatmap of all OTU relative abundances are available in Appendix A.ere were persistent OTUs in the T4-like myoviral community, but not in all of thephylogenetic groups (Figure 3.12). Some persistent OTUs remained even when phyloge-netically related OTUs were undetectable. In the marine picorna-like viruses, constantOTUs were oŸen phylogenetically similar to ephemeral OTUs (Figure 3.13).ere werealso clear changes in OTU composition from fall to winter, as was seen for the marinepicorna-like viruses (Figure 3.5 B), however some OTUs persisted. Nonetheless, ratherthan a gradual expansion of the OTUs, changes tended to occur quickly among groupsof related viruses.81chapter 3Figure 3.13: Plot of relative abundance of marine picorna-like OTUs (95% amino acidsimilarity) that were present in over 90% of samples (persistent) or in less than 20% ofsamples (ephemeral). OTUs are ordered by phylogenetic tree. Stars (*) indicate missingsamples. A heatmap of all OTU relative abundances are available in Appendix A.82chapter 33.4.8 Lagged correlations with hosts over timeRaphidophytes and marine picorna-like viral group A — e relative abun-dance of group A (Figure 3.5A), which includes HaRNAV, a virus which infects theraphidophyte Heterosigma akashiwo, increased in relative abundance aŸer the increasein relative abundance of eukaryotic sequences classied as raphidophytes (Figure 3.14,correlation: 0.43 and p value 0.06).ere were further peaks in raphidophytes, but theydid not coincide with an increase in the relative abundance of marine picorna-like viralgroup A.Of all the OTUs in group A, OTU 1 was the most relatively abundant and dišerencesin other sequences to it are highlighted (Figure 3.15), there were changes in an aminoacid from D to E in the palm region of the RdRp (te Velthuis, 2014).CyanobacteriaandT4-likemyoviralgroup I — Comparing theT4-likemyoviralgroup I, which contains cyanophage isolates, to the cyanobacterial OTUs (Figure 3.16),showed that the relative abundance of viruses increased in the fall aŸer peaks in thecyanobacterial OTUs (correlation: 0.5, p value: 0.03).e lags in relative abundances ofputative cyanophages relative to cyanobacteria continued, and aŸer the spring bloomthere was a lag before there was an increase in the relative abundance of a dišerentputative cyanophages in group I, showing succession in the viral community.Table 3.3: Spearman correlations among environmental parameters, communityrichness, and community similarity. Includes correlations that were lagged over time:each time point was compared to the previous timepoint (~2 weeks earlier). * is p < 0.05,SR is Species richness. (continued below)SR bacteria(16S)SReukaryotes(18S)Bray-curtis bacteria (16S) - -0.3883chapter 3Table 3.3: continuedSR bacteria(16S)SReukaryotes(18S)Bray-curtis bacteria (16S)lagged- -0.34Bray-curtis eukaryotes(18S) -0.37 -Bray-curtis eukaryotes (18S)lagged-0.51 -Bray-curtis marine picorna-like(RdRp)0.08 -0.09Bray-curtis marine picorna-like(RdRp) lagged-0.45 -0.15Bray-curtis T4-like myoviruses(gp23)0.13 0.08Bray-curtis T4-like myoviruses(gp23) lagged-0.08 -0.06SR marinepicorna-like(RdRp)SR T4-likemyoviruses(gp23)Bray-curtis bacteria (16S) - -0.18Bray-curtis bacteria (16S)lagged- -0.07Bray-curtis eukaryotes(18S) -0.07 -0.25Bray-curtis eukaryotes (18S)lagged0.33 -0.3684chapter 3SR marinepicorna-like(RdRp)SR T4-likemyoviruses(gp23)Bray-curtis marine picorna-like(RdRp)- -Bray-curtis marine picorna-like(RdRp) lagged- -Bray-curtis T4-like myoviruses(gp23)- -0.28Bray-curtis T4-like myoviruses(gp23) lagged- 0.13ViralabundanceBacterialabundanceBray-curtis bacteria (16S) - -0.24Bray-curtis bacteria (16S)lagged-0.02 -0.11Bray-curtis eukaryotes(18S) -0.19 -0.17Bray-curtis eukaryotes (18S)lagged-0.31 -0.11Bray-curtis marine picorna-like(RdRp)-0.19 -0.15Bray-curtis marine picorna-like(RdRp) lagged-0.22 0.08Bray-curtis T4-like myoviruses(gp23)0.16 -0.0685chapter 3ViralabundanceBacterialabundanceBray-curtis T4-like myoviruses(gp23) lagged0.12 -0.14Chlorophyll a PO4 SiO2NO3+NO2Bray-curtis bacteria (16S) -0.37 * 0.18 0.18 0.18Bray-curtis bacteria (16S)lagged0 0.02 0.06 0.05Bray-curtis eukaryotes(18S) -0.19 0.2 0.35 * 0.2Bray-curtis eukaryotes (18S)lagged0.17 0.24 0.18 0.16Bray-curtis marinepicorna-like (RdRp)-0.13 0.2 0.31 0.07Bray-curtis marinepicorna-like (RdRp) lagged0.01 0.24 0.17 0.26Bray-curtis T4-likemyoviruses (gp23)0.12 - -0.07 0.02Bray-curtis T4-likemyoviruses (gp23) lagged0.07 -0.01 -0.18 -Temperature SalinityDissolvedoxygen (%)Bray-curtis bacteria (16S) 0.07 -0.02 -0.286chapter 3Temperature SalinityDissolvedoxygen (%)Bray-curtis bacteria (16S)lagged-0.05 0.04 0.02Bray-curtis eukaryotes(18S) -0.16 0.17 -0.1Bray-curtis eukaryotes (18S)lagged-0.26 0.12 -0.06Bray-curtis marinepicorna-like (RdRp)-0.2 0.06 -0.06Bray-curtis marinepicorna-like (RdRp) lagged-0.32 0.1 -0.03Bray-curtis T4-like myoviruses(gp23)-0.15 0.09 -0.05Bray-curtis T4-like myoviruses(gp23) lagged0.01 0.1 -pHBray-curtis bacteria (16S) -0.11Bray-curtis bacteria (16S) lagged -0.05Bray-curtis eukaryotes(18S) -0.1Bray-curtis eukaryotes (18S)lagged-0.07Bray-curtis marine picorna-like(RdRp)-0.11Bray-curtis marine picorna-like(RdRp) lagged-0.3587chapter 3pHBray-curtis T4-like myoviruses(gp23)-0.02Bray-curtis T4-like myoviruses(gp23) lagged0.01Correlations between community similarity and richness — e communitysimilarity of the T4-like myoviruses had a strong lagged negative correlation to the rich-ness of the bacterial community (Table 3.3 and Table 3.9). e correlations to the MPLcommunity were much stronger to other communities when lagged than when directlycompared. Viral abundance was negatively correlated to bacterial community similarity.Table 3.9: Mantel tests among community similarity matrices and distance matrices ofenvironmental data. * is p < 0.05, ** p < 0.01Bacteria (16SrRNA gene)Eukaryotes(18S rRNAgene)Marinepicorna-like (RdRP)T4-likemy-oviruses(gp23)Viral abundance -0.1 -0.14 0.17 -0.15Bacterial abundance 0.02 -0.11 0.19 0.08Chlorophyll a -0.1 -0.19 -0.1 0.28PO4 -0.03 -0.04 0.07 0.06SiO2 0 -0.03 0.04 0.09NO3+ NO2 -0.05 -0.06 0.02 0.03Temperature 0.02 -0.03 0.15 -0.13Salinity -0.01 -0.08 0.17 0.03Dissolved oxygen (%) -0.1 -0.14 0 0.0288chapter 3Table 3.9: continuedBacteria (16SrRNA gene)Eukaryotes(18S rRNAgene)Marinepicorna-like (RdRP)T4-likemy-oviruses(gp23)pH -0.18 0.04 -0.06 -0.05Bacteria (16S rRNAgene)- 0.81 ** 0.52 ** 0.18 *Bacteria (16S rRNAgene) lagged- 0.4 ** 0.34 * 1 **Eukaryotes (18S rRNAgene)0.81 ** - 0.34 ** 0.3 **Eukaryotes (18S rRNAgene) lagged0.91 ** - 1 ** 0.94 **T4-like myoviruses(gp23)0.18 * 0.3 ** 0.47 ** -T4-like myoviruses(gp23) lagged0.01 0.37 ** 0.5 ** -Marine picorna-like(RdRP)0.52 ** 0.34 ** - 0.47 **Marine picorna-like(RdRp) lagged1 ** 0.37 ** - 0.97 **Mantel tests among community similarity — Mantel tests examined concurrentcommunity changes in distance matrices over time (Table 3.9). e bacterial and eu-karyotic community compositions žuctuated strongly together.e marine picornalikecommunity žuctuated with viral abundance and with salinity, more strongly with the89chapter 3Figure 3.14: Marine picorna-like viral group A compared to eukaryotic OTUs classiedas raphidophytes over time. A) Relative abundance of eukaryotic OTUs (97%) classiedas Raphidophyte (Stramenopiles) over time. B) Relative abundance of marine picorna-like virus group AOTUs (95% amino acid) over time. Each coloured contour representsa separate OTU. Grey vertical lines indicate boundaries between seasons and the greenvertical line indicates the time of the spring bloom.90chapter 3Figure 3.15: Marine picorna-like group A sequences aligned. Dišerences to the mostabundant OTU are highlighted.91chapter 3Figure 3.16: T4-like myoviral group I compared to bacterial OTUs classied ascyanobacteria over time. A) Relative abundance of bacterial OTUs (97%) classied ascyanobacteria over time. B) Relative abundance of T4-likemyovirus group I OTUs (95%amino acid) over time. Each coloured contour represents a separate OTU. Grey verticallines indicate boundaries between seasons and the green vertical line indicates the timeof the spring bloom.92chapter 3bacterial community, and slightly less strongly with the eukaryotic community. T4-likemyoviruses showed changeswith the eukaryotes andwith themarine picornalike viruses,but not with environmental parameters.3.5 discussion3.5.1 Major ndingsUsing high-throughput sequencing of samples collected at a coastal site every two weeksfor one year the dynamics of two groups of ecologically important groups of viruseswere described in the context of their putative hosts and the environment. Communitydynamics revealed dišerences by season. ere was a large diversity of viruses andputative hosts in this study, and groups of phylogenetically-related viruses showed tem-poral dynamics in dominance. While some members of these related groups persistedthroughout time, others were more ephemeral. ese ndings were put in context ofpotential quasispecies behaviour, of the dynamics of putative hosts, and of the seed bankand Killing the Winner theories.3.5.2 ShiŸs in dominance of viral communities by related viruses over timeAmajor theme in microbial ecology is describing temporal shiŸs in community compo-sition. Observing viral communities using the viral gene marker pho h, highly unevencommunities were found at Bermuda Atlantic Time-series Study (BATS) using deepamplicon sequencing (Goldsmith et al., 2015). Populations were divided into phylo-genetically distinct groups and dišerences between the fall and winter samples wereattributable to a phylogenetic group containing cyanobacterial infecting viruses and towater stratication (Goldsmith et al., 2015). In this study at Jericho Pier, even thoughthere was a large diversity of OTUs in the viral communities, the overall temporal dy-namics were driven by shiŸs in phylogenetically-related OTUs (Figure 3.5). e MPLviruses had similar dynamics between the related viruses (Figure 3.5B) and the top 2093chapter 3MPL OTUs over time (Figure 3.8) indicating an uneven community as previously seen(Culley et al., 2006; Gustavsen et al., 2014). Conversely, in the T4-like myoviruses thephylogenetically related group patterns did not resemble the patterns of the top 20OTUsindicating that the communitywasmuchmore diverse and even. Phylogenetic dynamicsare important and žuctuate even though viral richness žuctuatedminimally (Figure 3.4).e dynamics of the phylogenetic groups are in agreement with Rodriguez-Britoet al. (2010) where the viral and microbial communities were stable at the genus andabove taxonomic levels, but were dynamic at the strain or species level. Furthermore,taxonomically related species can have more similar niches and ecology (Harvey andPurvis, 1991; Srivastava et al., 2012), thus it is possible these related viruses could beinfecting similar hosts or responding to environmental cues in similar ways or both.3.5.3 Groups of related viruses contain ephemeral OTUs and constant OTUs (quasis-pecies)In Figure 12 and Figure 13 there are OTUs that are persistent and thus continually suc-cessful, and others that are ephemeral. In the T4-like myoviruses there are phylogeneticgroups that do not contain persistent OTUs, but still have ephemeral OTUs that pe-riodically dominate the viral community (Figure 3.12). Rodriguez-Brito et al. (2010)found that the largest OTUs were persistent throughout time and suggested that tran-sient viruses were washed in from dišerent areas. is seems unlikely for this studysince the ephemeral viruses at Jericho Pier were closely related to the persistent viruses.Furthermore, both persistent and ephemeral viruses of phytoplankton were found ina freshwater lake (Short and Short, 2009; Rozon and Short, 2013), and at a coastal site(Short and Suttle, 2003) where some ephemeral viruses were correlated with shiŸs inenvironmental parameters. Additionally, inmarine T4-likemyoviral communities someOTUs were persistent and many more were ephemeral in 3-year (Chow and Fuhrman,2012) and 2-year (Pagarete et al., 2013) time series. erefore, both eukaryotic and bac-terial marine viruses show this structure of ephemeral and persistent viruses.94chapter 3e population structure of RNA viruses is proposed to be a mixture of genotypescalled quasispecies that are produced through errors and encompass the communityof genotypes theoretically produced from one infection (Holmes, 2010; Domingo et al.,2012). us a sequenced viral genome is oŸen an “average” of all of these individualgenotypes. In an Antarctic lake the ecological setting likely inžuences the presence ofquasispecies in viral RNAmetagenomes (López-Bueno et al., 2015) since the number ofquasispecies recovered from in the lake water compared to the microbial mats over timewas very dišerent.ere weremore single nucleotide variations (SNVs) in the lakewatermetagenomes since either there is more turnover and more ecological niches/diversityin these samples, or it is just the result of convergence of water from more locations.Quasispecies behaviour has been best characterized inRNAviruseswhich are knownto have a higher mutation rate and have higher burst sizes compared to the dsDNAviruses (Milo et al., 2010). However, it is theoretically possible for quasispecies to ex-ist in bacteriophage populations (Weitz et al., 2005). In high condence viral DNAmetagenomes there is heterogeneity in assembled reads beyond just expected sequenc-ing errors (Dutilh et al., 2014), and there is site-specic variation in genomes fromDNAviral populations studied in humans (Renzette et al., 2015). As was found in López-Bueno et al. (2015), Renzette et al. (2015) saw that certain regions of the genomes hadmore SNVs than others indicating that selection does not appear constant across thegenomes. When the OTUs in the marine picorna-like group A are compared in analignment (Figure 3.15) many of the mutations in the ephemeral OTUs are randomlyspread in the gene fragments, but most of them change from D to E in the “palm” sec-tion of the catalytic site C (te Velthuis, 2014). is would point to a population thatis marginally successful while the most abundant OTU (retaining the D amino acid)remains persistent. is raises questions about whether this phenomenon is prevalentin marine settings and should be incorporated into current ecological theories (e.g. seedbank and KtW).95chapter 33.5.4 Implications for theories related to community structure and dynamicsIn the Killing the Winner theory viruses infect the most active organism (ingstad,2000). Hosts compete for limiting resources which determines composition by site andviruses determine the identity and abundance of these hosts (Storesund et al., 2015).With the lagged dynamics in the raphidophyte and viruses related to raphidophyte infect-ing viruses, the “winner” has been killed (Figure 3.14).ere is a later small increase inrelative abundance of the Raphidophytes with no associated or lagged increase in RdRpviral groupA.One explanation is that the number of susceptible hosts were not abundantenough for a detectable increase in the viral group A, or the host could be targeted bya dišerent subset of viruses (e.g. HAV, a DNA virus, also infects HAKA (Nagasaki andYamaguchi, 1997)), or the hosts could be being controlled by another protist (ciliate e.g.(Harvey and Menden-Deuer, 2012)). Similar patterns were detected in the cyanophagefrom group I and the cyanobacteria (Figure 3.16). ese types of patterns have beenseen in (Chow et al., 2014) where there were many viral OTUs detected with strong time-lagged correlations to bacterial OTUs. Also, in a mesocosm experiment with Emilianiahuxleyi there was a peak in host abundance and then four days later a peak in Emilianiahuxleyi Virus (EhV) abundance (Schroeder et al., 2003). Rapid shiŸs in ne-scale viraldynamics and stability at coarse scale viral dynamics suggest that the Killing theWinnertheory is operating at the strain level (Rodriguez-Brito et al., 2010; Emerson et al., 2013)preserving bacterial strain level diversity (Rodriguez-Valera et al., 2009;ingstad et al.,2015).In the marine environment, viruses, at a coarse scale of family/genus, could showKilling the Winner dynamics, and then at a ner scale could form a “seed bank” wherethere is shuœing of phylogenetically-related viruses on a rank abundance curve. isbank or seed bank model (Breitbart and Rohwer, 2005) which explains how high localviral diversity (shuœing of viruses) can be consistent with low overall global diversity(most abundant viruses) by a constant local production of viruses has been supportedby many studies (Short et al., 2010; Chow and Fuhrman, 2012; Zhong and Jacquet, 2014;96chapter 3Brum et al., 2015). Our results are concordant with Goldsmith et al. (2015) who foundthat the viral community was mostly dominated (over 50%) by a few successful OTUsand the rest of the OTUs were rare and contained in the “bank”. A further layer hasbeen added to this idea by showing that the relatedness of the viruses in this seed bankis crucial to understanding their dynamics. Our data reveals that the viruses in theseed bank can be ephemeral and related to persistent viruses and that groups of relatedviruses can become abundant through ecological processes (through habitat ltering e.g.Koeppel and Wu (2013a)).3.5.5 CaveatsAlthough the challenges with viral gene markers has been previously discussed (Gus-tavsen et al., 2014) and PCR in general (Lee et al., 2012), this is a useful approach forexamining the population structures. Furthermore, especially with the rare and putativequasispecies OTUs, it cannot be ruled out that they are erroneous due to PCR (Pinto andRaskin, 2012) or sampling anomalies (as discussed in Shade et al., 2014), but these OTUswere seen multiple times in dišerent samples therefore seem less likely to be spurious.To increase condence in the results, some libraries with lower numbers of reads wereexcluded so that more sequence could be used overall when normalizing samples (Pintoand Raskin, 2012).e sequences were checked for chimeras since chimeras can form asa result of high cycle number as used one of the targets (Qiu et al., 2001). Read abundanceof OTUs can be considered semi-quantitative and good for comparisons of richness anddiversity among samples (but not for absolute counts of genes) (Pinto and Raskin, 2012).3.5.6 ConclusionsRelated viruses show temporal patterns of dominance over time. ere were stronglagged correlations between hosts and groups of related viruses and viral communitiesshow evidence of following host communities. More ne-scale structure of the commu-nities was dependent on the dišerent life strategies of the viral communities examined.97chapter 3e marine picorna-like viruses exhibited more quasispecies-type behaviour and theT4-like myoviruses viruses, with their theoretically lower burst sizes, appeared to havea dišerent mix of persistent and ephemeral viral dynamics. Overall, viral communitydynamics are largely inžuenced by phylogenetically related groups of viruses over timeand these dynamics add another layer to the well-established theories of Killing theWinner and the seed bank model for viral communities.98chapter 4Network analysis of Jericho Pier microbial time-series4.1 summaryMarine microbes play a fundamental role in the dynamics of the marine ecosystem.eco-occurrence between these microbes can show links between organisms or sharedniches, and thus describes the structure, dynamics and stability of these communities.Microbial co-occurrence networks at global ocean and coastal ocean scales have foundexpected, time-lagged and new links. Yet, no studies have included multiple groups ofviruses in the association networks. To investigate the ecological patterns and driversof diversity in microbial communities, high-throughput sequencing was performed foreukaryotic (18S ribosomal RNA gene), bacterial (16S ribosomal RNA gene) and viral(gp23 for T4-like myoviruses and RNA dependent RNA polymerase (RdRp) for ma-rine picorna-like viruses) amplicons from a 1-year time series at a coastal site in BritishColumbia, Canada. Using local similarity analysis (LSA), co-occurrence networks andthe phylogenetic relatedness in these communities were examined. e network topol-ogy revealed that within the viral communities there were more links than within thebacterial and eukaryotic communities, and within the viral networks phylogenetically-related viruses separated into tightly-connected subnetworks (modules). Examiningthe co-occurrence of operational taxonomic units (OTUs) over time, the T4-like my-oviral community had the greatest number of links that were strongest when they werecompared to the previous time-point. Over time, the eukaryotic and bacterial OTUswere more strongly correlated to environmental factors than the viral OTUs, thus thesecommunities were strongly driven by the environment. Communities sampled in the99chapter 4fall were more strongly correlated than in other seasons and shared the greatest numberof links with the winter timepoints, indicating a time of stability for these communities.us, the ecological interpretations ofmicrobial association networks, with the inclusionof viruses, could further our understanding of the drivers of microbial diversity andassembly.4.2 introductione high diversity, high abundance, and dynamics of marine microbial communitieshave begun to be explored, largely through advances in sequencing.ough oŸen over-looked, it is important to integrate viruses into these community analyses since virusescanhave a large inžuence onbacterial and eukaryotic communities. For example, virusesare estimated to be responsible for the daily lysis of 10-50% of heterotrophic bacteriaand 5-10% of the cyanobacteria (Wilhelm and Suttle, 1999; Weinbauer, 2004) in plank-ton. Some studies have found that viruses control the host population compositionor abundance or both (Bouvier and del Giorgio, 2007; Storesund et al., 2015). Usingmicrobial association networks, Chow et al. (2014) found that viruses may follow theirhost’s abundance rather than control it. Moreover, in a global ocean survey of surfacewaters, viruses were found to be host-range limited because of large geographic distances(Lima-Mendez et al., 2015).An important step in understanding why and how marine bacterial, protistan andviral communities show seasonality and repeatability over time in composition and abun-dance (Gilbert et al., 2011; Chow and Fuhrman, 2012; Chow et al., 2013; Fujiki et al., 2014;Simon et al., 2015) is to explore the co-occurrences of individual taxa in communitiesover time, and how these relationships are ašected by environmental and biotic factors.Characterizing these relationships can also help towards incorporating the role of virusesin maintaining the diversity of host populations and for understanding the dynamics ofthese systems.100chapter 44.2.1 Network analysis to examine relationshipsCo-occurrence of organisms over time can be used to infer relationships or sharedniches;these co-occurrences can be visualized using network diagrams. Network analysis hasbeen used to examine relationships inmicrobial communities that are otherwise di›cultto visualize, to examine known relationships, to propose new putative relationships,and to examine the overall structure of communities (Faust et al., 2015b; Lima-Mendezet al., 2015). Network analysis is used to examine association matrices produced bypairwise distance matrices (e.g. Bray-Curtis), Spearman or Pearson correlations, localsimilarity analysis (Ruan et al., 2006; Xia et al., 2011), generalized boosted linear models(Faust and Raes, 2012) or other association measures. Associations can be positive ornegative, and can be the result of symbiosis (complementary functions), similar niches,competition, dišerent resource use, predation, viral lysis, or grazing (ideas fromSchluter,1984; Chow et al., 2014). Network analysis can be used to examine co-occurrence oforganismswithout knowing if the relationships are direct (e.g. predation or viral lysis) orindirect (e.g. density-mediated interactions) (Miki and Jacquet, 2010). As well, detectionof microbial keystone species, those crucial for the overall ecosystem or community,shows promise by the examination of topological network characteristics (number ofconnections, and network properties when organism is removed) (e.g. Berry and Wid-der, 2014; Williams et al., 2014).Networks have also been used to examine the niches of organisms and be used toreveal the niches occupied by microbes (Steele et al., 2011). Network analysis has alsobeen used to assess temporal associations in marine systems (Gilbert et al., 2011; Chowet al., 2013, 2014; Cram et al., 2015), and spatial associations in oceans (Lima-Mendezet al., 2015), soils (Barberán et al., 2012), and permafrost thaw ponds (Comte et al., 2015).Time series studies at the San Pedro Ocean Time series (SPOT) in California have foundrepeatable patterns in the bacterial and viral communities, and strong connections inmicrobial networks illustrating known and new putative interactions (Chow et al., 2014).Detecting such groups of highly-connected organisms can allow inferences to be made101chapter 4with respect to temporal or seasonal variability (Cram et al., 2015). Aswell, co-occurrencenetworks can help identify potential host-virus pairs and illuminate the potential ecol-ogy of these viruses. For example, viral and bacterial networks constructed from a globalocean survey showed that 43% of the phage populations were only strongly correlated toone bacterial OTU and the rest (57%) were strongly correlated to a few bacterial OTUs(Lima-Mendez et al., 2015).Microbes, including viruses, are key members of marine ecosystems and play largeroles in geochemical cycling, photosynthesis, and nutrient remineralization (Wordenet al., 2015). Despite seasonal forcing, communities exhibit resilience implying that in-ternal factors, such as composition or diversitymay account for this resilience (Fuhrmanet al., 2015; Faust et al., 2015a). Previous network analysis of time-series data formicrobialcommunities provided important insights, but generally did not examine phylogeneticrelationshipswithin these communities, norwhich co-occurrencesweremost importantfor dening the communities temporally. Phylogeny is important because co-occurringmicrobes can be phylogenetically closely related (Chašron et al., 2010). Using networksto analyze the dynamics of marine microbial communities can reveal how these systemsrespond to change over time, as well as how they maintain resilience and stability.4.2.2 ApproachUsing marker genes for two group of viruses (the T4-like myoviruses and the marinepicorna-like viruses, as described in the Introduction and Chapter 3), bacteria, andeukaryotes, these communities were examined in a one year time series with samplestaken every two weeks at Jericho Pier, in Vancouver, British Columbia. Local similarityanalysis (LSA) was performed on the operational taxonomic units (OTU) and environ-mental parameters to examine the co-occurrence of all possible pairs within the sametime-point (no lags) and with lags of up to one month (two timepoints). Additionally,to test the inžuence of the environment on these communities, redundancy analyses(RDA) and variation partitioningwere performedwith the community data and environ-102chapter 4mental parameters. e aim of collecting these data are to determine the relationshipsof the viral, bacterial, and eukaryotic communities over time and the inžuence of theenvironment on driving the changes in these communities.4.2.3 HypothesesPatterns of co-occurrence Viral replication depends on infection; hence, viruses mustco-occur with their hosts either at the same time or with time-lags. erefore, strongcorrelations in co-occurrence should occur between a virus and its host, and potentiallybetween viruses and other organisms. However, dišerent patterns in co-occurrencewould be expected given the range of lifestyles of hosts and their viruses. For example,T4-like myoviruses can have narrow or broad host ranges (Sullivan et al., 2003) andburst sizes of about 25 to 200; whereas, marine picornalike viruses are host-specic andhave burst sizes > 1000 (Lang et al., 2009). Viruses with high host specicity wouldproduce networks with fewer connections to putative hosts (but does not limit intra-virus connections) than viruses with a wider host range, such as in some of the T4-likemyoviruses.Temporal niches.e environment inžuences viral diversity beyond regulating hostdiversity; for example, temperature and salinity can ašect the ability of viruses to infecthosts (Kendrick et al., 2014). RDA will be used with community-level data to infer whatis driving the diversity of dišerent communities. To observe another aspect of temporalniches, at the OTU-level, subnetworks, composed of a strongly correlated OTU pair anda strongly correlated environmental parameter (environmental triplets), will be used toexamine environmental drivers of OTU pairs.Resilience and stability Most microbial networks have “small-world” properties,meaning that the number of nodes separating two organisms is low, even with a highnumber of organisms, which suggests that these communities should be resistant tochange. Hence, it is hypothesized that the microbial networks will also have few con-nections between organisms.103chapter 44.3 materials and methods4.3.1 Sample collectionSampleswere collected from JerichoPier (49° 16’36.73N, 123° 12’05.41W) inBritishColumbia,Canada. Jericho Pier (JP) is adjacent to the shoreline, in a well-mixed location withmixed semi-diurnal tides. Sixty litres of water were pumped from 1m depth every twoweeks at the daytime high tide between June 2010 and July 2011, inclusive. Salinity, tem-perature and dissolved oxygen were measured using a YSI probe (Yellow Springs, Ohio,USA). For all samples, the water was ltered sequentially through 142-mm diameter, 1.2µmnominal pore-size glass-bre (GC50AdvantecMFS,Dublin, CA., USA) and0.22 µmpore-size polyvinyldine lters (Millipore, Bedford, MA, USA).e ltrate, containingthe viral size fraction, was concentrated to ~500 mL (viral concentrate) using tangentialžow ultraltration with a 30kDa MW prep-scale Spiral Wound TFF-6 cartridge (Milli-pore) (Suttle et al., 1991).Phosphate, silicate, and nitrate+nitrite concentrations were determined in duplicate15 mL seawater samples ltered through 0.45 µm pore-size HA lters (Millipore) andstored at -20°Cuntil air-segmented continuous-žowanalysis on aAutoAnalyzer 3 (Bran+Luebbe, Norderstedt, Germany). Chlorophyll a (Chl a) was determined in triplicate byltering 100mL of seawater through 0.45 µmpore-size HA lters (Millipore).e lterswere stored in the dark at -20°C until acetone extraction and then analysed žuorometri-cally (Parsons et al., 1984).4.3.2 Enumeration of bacteria and virusesSamples for viral and bacterial abundances were taken at each sampling point by x-ing duplicate cryovials containing 980 µL of sample with nal concentration of 0.5%glutaraldehyde (EM-grade, EMS, Hateld, PA, USA), freezing in liquid nitrogen andstoring at -80°C until processing. Briežy, viral samples were diluted 1:10 to 1:10 000in sterile 0.1 µm ltered 1X TE, stained with SYBR Green I (Invitrogen, Waltham, MA,104chapter 4USA) at a nal concentration of 0.5 x 10-4 of commercial stock, heated for 10min at 80° Cand then cooled in the dark for 5 min before processing. Bacterial samples were dilutedup to 1:1000 in sterile 0.1 µm ltered 1X TE, stained with SYBR Green I (Invitrogen)at a nal concentration of 0.5 x 10-4 of commercial stock, and incubated in the darkfor 15 min before processing. All samples were processed on a FACScalibur (Becton-Dickinson, Franklin Lakes, New Jersey, USA) with viral and bacterial samples run for1 min at a medium or high žow rate, respectively. Event rates were kept between 100to 1000 events per second and green žuorescence and side scatter detectors were used.Data were processed and gated using Cell-Quest soŸware (Becton-Dickinson).4.3.3 Viral concentration and extractione viral concentrate was ltered twice through 0.22 µm pore-size Durapore PVDFlters (Millipore) in a sterile Sterivex lter unit (Millipore).e ltrate, containing virus-sized particles, was pelleted by ultracentrifugation (Beckman-Coulter, Brea, California,USA) in a SW40 rotor at 108 000 g for 5 h at 12°C.e pellet was resuspended overnightin 100 µL of supernatant at 4°C. To digest free DNA, the pellets were incubated with1U/µL DNAse with a nal concentration 5 mM MgCl2 for 3 h at room temperature.Nucleic acids were extracted using a Qiamp Viral Minelute spin kit (Qiagen, Hilden,Germany) according to the manufacturer’s directions.4.3.4 PCR amplication of T4-like myoviral marker geneTo target themarine T4-type bacteriophage capsid protein gene (gp23) PCRs were set upas in Filée et al. (2005). Briežy, each reaction mixture (nal volume, 50 µL) consisted of2 µL template DNA (approx. 40 ng/µL measured with a Nanodrop (ermo Fisher Sci-entic, Waltham, MA, USA)), 1x (nal concentration) PCR bušer (Invitrogen), 1.5 mMMgCl2, 0.2 mM of each deoxynucleoside triphosphate (Bioline, London, UK), 40 pmolof MZIA1bis, 40pmol of MZIA6, and 1 U Platinum Taq DNA polymerase (Invitrogen).Program conditions as in Table 3.1.105chapter 44.3.5 PCR amplication of marine picorna-like viral marker geneHalf of the viral extract was used for the generation of cDNA. To remove DNA, theextracted viral pellets were digested with DNase 1 (amplication grade) (Invitrogen).e reaction was terminated by adding 2.5 mMEDTA (nal concentration) and incubat-ing for 10 min at 65°C. Complementary DNA (cDNA) was generated using SuperscriptIII Reverse Transcriptase (Invitrogen) with random hexamers (50 ng/µL) as per themanufacturer.PCR was performed with primer set MPL-2 for a targeted set of the marine picorna-like virus RdRp (Culley and Steward, 2007). Each reaction mixture (nal volume, 50µL) consisted of 50 ng of cDNA, 1x (nal concentration) PCR bušer (Invitrogen), 2mM MgCl2, 0.2 mM of each deoxynucleoside triphosphate (Bioline, London, UK), 1µM of each primer, and 1 U Platinum Taq DNA polymerase. e reaction was runin a PCR Express thermocycler (Hybaid, Ashford, UK) with program conditions as inTable 3.1. Products were run on a 0.5X TBE 1% lowmelt gel, excised and extracted usingZymoclean Gel DNA Recovery Kit (Zymo) as per manufacturer’s directions with a nalelution step of 2x10 µL EB bušer (Qiagen).4.3.6 Filtration and extraction of marine bacteria and eukaryotesOne liter of seawater was ltered through a 0.22 µm pore-size Durapore PVDF 47 mmdiameter lter (Millipore) in a sterile Sterivex lter unit (Millipore). e lter was ei-ther immediately extracted or was stored at -20°C until extraction. Filter extractionproceeded as in Short and Suttle (2003). Briežy, lters were aseptically cut and incubatedwith lysozyme (Sigma-Aldrich, St. Louis, MO, United States) at a nal concentration of 1mgmL-1 for 2 h at 37°C. Sodium dodecyl sulfate was added at a nal concentration of 0.1% (w/v) and each lter was put through three freeze-thaw cycles. Proteinase K (Qiagen)was then added to a nal concentration of 100 µg/mL and incubated for 1 h at 55°C. DNAwas sequentially extracted using equal volumes of phenol:chloroform:IAA (25:24:1), andchloroform:IAA (24:1). DNAwas precipitated by addingNaCl to a nal concentration of106chapter 40.3M and by adding 2X the extract volume of ethanol. Samples were incubated at -20°Cfor at least 1 h and then centrifuged for 1 h at 20 000 g at 4°C. Extracts were washed with70 % ethanol and then aŸer drying were resuspended in 50 µL EB bušer.4.3.7 PCR amplication of marine bacteria and eukaryotesPCR targeting the eukaryotic fraction of the extract was performed using Euk1209f andUni1392r primers as in Diez et al. (2001). Briežy, each reaction mixture (nal volume,50 µL) consisted of 2 µL template, 1x (nal concentration) PCR bušer (Invitrogen), 1.5mM MgCl2, 0.2 mM of each deoxynucleoside triphosphate (Bioline, London, UK), 0.3µMof each primer, and 2.5 U Platinum Taq DNA polymerase.e reaction was run in aPCR Express thermocycler (Hybaid, Ashford, UK) with program conditions as in Table3.1.PCR targeting the bacterial fractions using primers 341F (Baker et al., 2003) and 907R(Muyzer et al., 1995)was performedwith the following conditions: each reactionmixture(nal volume, 50 µL) consisted of 2 µL template, 1x (nal concentration) PCR bušer(Invitrogen), 1.5 mM MgCl2, 0.2 mM of each deoxynucleoside triphosphate (Bioline,London, UK), 0.4 µM of each primer, and 1 U Platinum Taq DNA polymerase. ereaction was run in a PCR Express thermocycler (Hybaid, Ashford, UK) with programconditions as in Table 3.1.4.3.8 Sequencing library preparationConstruction — PCR products not requiring gel excision were puried using AM-Pure XP beads (Beckman Coulter) at a ratio of 1.2:1 beads:product. Cleaned productswere resuspended in 30 µL EB bušer (Qiagen). All products were quantied usingPicogreen dsDNA (Invitrogen) assay using Lambda DNA (Invitrogen) as a standard.Sample concentrations were read using iQ5 (Bio-Rad, Hercules, CA, USA) and CFX96Touch systems (Bio-Rad). Pooled libraries were constructed using one of each of theamplicons so that their molarity would be similar and totalling ~700-900 ng. Pooled107chapter 4amplicons were concentrated using AMPure XP beads (Beckman Coulter) at a ratio of1.2:1 beads:product. NxSeq DNA sample prep kit 2 (Lucigen, Middleton, WI, USA) wasused as per manufacturer’s directions with either NEXTžex 48 barcodes (BioO, Austin,USA), NEXTžex 96 HT barcodes (BioO), or TruSeq adapters (IDT, Coralville, Iowa,USA). Small Libraries were cleaned up using AMPure XP beads (Beckman Coulter) ata ratio of 0.9:1 beads:library.Quantification and quality control of libraries — Libraries were checked forsmall fragments (primer dimers and/or adapter dimers) using a 2100 Bioanalyzer (Agi-lent, Santa Clara, CA, USA) with the High Sensitivity DNA kit (Agilent).e concentra-tion of libraries was quantied using Picogreen dsDNAassay as above.e libraries werechecked for quantication and for ampliable adapters using the Library QuanticationDNA standards 1-6 (Kappa Biosystems, Wilmington, USA) with the SsoFast EvaGreenqPCR supermix (Bio-Rad) using 10 µL EvaGreen master mix, 3 µL of 0.5 µM F primer,3 µL of 0.5um R primer and 4 µL of 1:1000, 1:5000 and 1:10000 dilutions of the librariesin triplicate on iQ5 (Bio-Rad) and CFX96 Touch qPCR machines. Cycling parameterswere as follows: 95°C for 30s, 35 cycles of 95°C for 5s, 60°C for 30s, and the melt curvegeneration from 65°C to 95°C in 0.5°C steps (10s/step). Quantication values obtainedfrom both Picogreen assays and qPCR assays were used to determine nal pooling of alllibraries before sequencing. Libraries were sequenced using 2 x 250bp PE Miseq (Illu-mina, San Diego, USA) sequencing at Génome Québec Innovation Centre at theMcGillUniversity (Montreal, QC, Canada), and 2 x 300bp PE Miseq (Illumina, San Diego, CA,USA) sequencing at UBC Pharmaceutical Sciences Sequencing Centre (Vancouver, BC,Canada) and at UCLA’s Genoseq (Los Angeles, CA, USA).4.3.9 Initial sequence processingLibraries were either split by adapter by the sequencing centre using CASAVA (Illumina)or by the user using the Miseq Reporter soŸware (Illumina). Sequence quality was108chapter 4initially examined using FastQC (Andrews, 2015). Contaminating sequencing adapterswere removed using Trimmomatic version 0.32 (Bolger et al., 2014) and sequencinglibrary quality was further examined using fastx_quality (Gordon, 2014). Libraries werefurther split into individual amplicons (i.e. 18S, 16S, gp23 and RdRp) and then, if the ex-pected overlap of the paired-end reads was 40bp or more, the paired reads were mergedusing PEAR version 0.9.6 (Zhang et al., 2014). Sequences were then quality trimmedusing Trimmomatic with the default quality settings. Reads were annotated by libraryand then all libraries from individual primers were added together. Sequences werealigned to known sequences (Silva 119 database (Quast et al., 2013) for 16S and 18S andfor viruses to alignments built from viral isolates and Sanger sequenced environmentalsurveys) using align.seqs in mothur 1.33.3 (Schloss et al., 2009). Reads not aligning wereremoved. Sequences were queried using BLAST against databases containing the genemarkers of interest downloaded from Genbank and sequences with an e-value below10-3 were kept.4.3.10 Chimera checking, OTU picking and read normalizatione 16S and 18S rRNA gene sequences were checked for chimeras using USEARCHversion 8.0.1517 reference (Edgar, 2010) with the Gold reference database. Unique, non-chimeric sequenceswere clustered at 97% similarity. Taxonomy for the 16S and 18S rRNAgene sequences was assigned usingmothur (Wang-type algorithm) and the taxonomy inSilva 119 (Quast et al., 2013). For the viral targets sequences were chimera-checked usingUSEARCH denovo and reference (Edgar, 2010). Viral sequences were then translatedusing FragGeneScan 1.20 (Rho et al., 2010). Viral reads were clustered using USEARCH(Edgar, 2010) at 95 % similarity for MPL, and 95 % similarity for T4-like myoviruses.Operational taxonomic unit (OTU) tables for all targets were constructed using USE-ARCH (Edgar, 2010). Rarefaction curves were generated using vegan (Oksanen et al.,2015). Sequences were normalized for this project by date and by target using vegan(Oksanen et al., 2015).109chapter 44.3.11 Data analysis and multivariate statisticsEnvironmental data — Environmental parameters were mean imputed to ll indata missing because of instrument malfunction or unavailability. Day length data wereretrieved using R package geosphere (Hijmans, 2015), and irradiance values from theUBCMeasurement Network (Christen, 2013).Hypothesis testing: RDA and Variation partitioning — Redundancy analysis(RDA) was performed on Hellinger transformed OTU tables to test whether the envi-ronmental parameters inžuenced the variability and structure of the communities overtime (Legendre and Legendre, 1998) in vegan (Oksanen et al., 2015). Highly co-relatedvariables were removed prior to starting the model. Signicance of the overall RDA wasdetermined using function anova.cca with 999 permutations and alpha of 0.05. For eachaxis the same check for signicance was performed using function anova.cca. Variationpartitioning (Borcard et al., 1992) was performed to nd the percentage of communityvariation determined from abiotic vs. biotic factors. Only signicant fractions were keptin the model.4.3.12 Network analysisLocal similarity analysis and network visualization — OTUs present in atleast 30% of sites were used for local similarity analysis (LSA). LSA was run for 29 timepoints with a maximum lag of two timepoints (equivalent to about a one month lag). Qvalues, used to evaluate the false discovery rate, were determined (Storey et al., 2005) andLSA resultswith aQ value less than 0.05 and p value less than 0.05were kept as signicantlinks. LSAs were formatted into networks and network statistics calculated using igraph(Csardi and Nepusz, 2006) in R and then visualized in Cytoscape 3.3.0 (Smoot et al.,2011). Cyrest (Ono et al., 2015) was used to send the networks to Cytoscape from R.Force-directed layouts were used from AllegroLayout 2.2.2 (Allegro, Santa Clara, CA,110chapter 4USA) in Cytoscape. Sub-networks were constructed by ltering the overall network byamplicon.To assess whether the networks had specic properties not only stemming fromthe number of nodes and edges, LSA networks were compared to simulated randomnetworks. Simulated networks were constructed using a random network generator(erdos.renyi.game) and scale-free network generator (barabasi.game) in igraph (Csardiand Nepusz, 2006).Module picking — Modules, which represent groups of highly connectedOTUs (sub-graphs), were determined using the cluster walktrap algorithm in igraph (Csardi andNepusz, 2006). is algorithm looks for highly connected nodes in a network usingrandom walks through the network. Walks with short path lengths indicate that nodeswere in the same community.Environmental triplets — To examine connections between OTUs driven by theenvironment, connections were examined where there were two correlated OTUs thatwere also each correlated to an environmental parameter giving “environmental triplets”4.4 results4.4.1 Co-occurrence of OTUs over 1 year-time seriesAs seen in Chapter 3, there was high overall diversity and richness in the OTUs in theviral, bacterial and eukaryotic networks. Using local similarity analysis (LSA) (Ruanet al., 2006; Xia et al., 2011) association networks were created from the signicant pair-wise associations. Sub-networks containing either only viral, bacterial, eukaryotic or amixture of nodes were created by ltering the overall network for specic types of OTUsand visualizing the connections.111chapter 4Figure 4.1: Legend for networks. Taxonomic phylum classications for bacteria andeukaryotes based on Silva 119 (Quast et al., 2013). Node shapes for viral OTUs andenvironmental parameters. Edge line type displays the delay between time pointswhere strongest correlation is found. Edge correlation type displays whether OTUs arenegatively or positively correlated. Colour of node outline by phylogenetic grouping forviral OTUs. Colours for these groups are from Chapter 3.112chapter 4Figure 4.2: Bacterial and eukaryotic networks. A) Bacterial OTU network colouredby phylum. B) Bacterial OTU network with modules. C) Eukaryotic OTU networkcoloured by phylum. D) Eukaryotic OTU network with modules. For legend see Figure4.1. Modules detected using the cluster walktrap algorithm. Members of the samemodule are the same colour. Node size is scaled based on degree, i.e. the number ofstrong co-occurences with that node.113chapter 4Community detection in networks — In the bacterial community network mod-ules, highly connected groups of OTUs in individual sub-networks, were visualized andthis highlighted two large, negatively correlated,modules (Figure 4.2B).Most of the linksbetween the two modules represented negative correlations among the Proteobacteriaand the Flavobacteria.e twomodules had similar composition, but one featuredmorehighly connected Actinobacteria (Figure 4.2A).e most well connected OTU in theoverall network was classied as Dežuviicoccus spp. (S16_127, Rhodospiralles, a class ofAlphaproteobacteria).e eukaryotic networks displayed similar patterns as the bacterial communitieswith two large groups of negatively correlated modules (Figure 4.2D).ese two groupshad similar composition and were largely connected by one 18S OTU (18S-3) classiedas a dinožagellate (class Gymnodiniphycidae) (Figure 4.2C).In the T4-like myoviral networks, the groups detected using the module algorithmwere very similar to the phylogenetic groups detected in Chapter 3 when overlaid ontothe network (Figure 4.3A and B).e most interconnected group of T4-like myovirusesshowedmany negative correlations to one of the modules that contained T4-like virusesfrom many dišerent phylogenetic groups.As with the T4-likemyoviral communities, themarine picorna-like viral communitycontained modules that were very similar to the phylogenetic groupings from Chapter 3(Figure 4.3C andD).ere were strong negative correlations between the largestmoduleand one of the other modules.In all the networks the links within modules were more oŸen positive than negativeand connections between modules were more oŸen negative.Co-occurrence between two types of amplicons — e association network forboth the bacterial and eukaryotic communities contained two main modules that werenegatively correlated (Figure 4.4A and B). Overall in this network the node classiedas Rhodobacteraceae (16S OTU-323) had the most connections to other OTUs. eassociation network for the bacterial community and the T4-like myoviruses contained114chapter 4Figure 4.3: Viral community networks. A) T4-like myoviral OTU network colouredby phylogenetic grouping. B) T4-like myoviral OTU network with modules. C) Marinepicorna-like OTU network coloured by phylogenetic grouping. D) Marine picorna-likeOTU network with modules. For legend see Figure 4.1. Modules detected using thecluster walktrap algorithm. Members of the same module are the same colour. Nodesize is scaled based on degree, i.e. the number of strong co-occurences with that node.115chapter 4Figure 4.4: Networks between dišerent communities. Bacterial and eukaryotic OTUnetwork A) coloured by phylum and B) modules coloured. Bacterial and T4-likemyoviralOTUnetworkC) coloured by phylumandD) coloured bymodules. Eukaryoticand marine picorna-like OTU network E) coloured by phylum and F) coloured bymodules. Legend Figure 4.1. Modules detected using the cluster walktrap algorithm.Members of the same module are the same colour. Node size is scaled based on degree.116chapter 4two main modules (Figure 4.4C and D). In one tightly clustered module there weremostly the phyla Alphaproteobacteria (Rhodobactierales, Rhodospirales, SAR 11), Be-taproteobacteria (Burkoholderiales, Flavobacteria, some other Bacteroides, Chlorožexi(SAR 202 clade), Deferribacteria), Deltaproteobacteria (Desulfobacteriales, SAR 324)Firmicutes (Clostridia), and Gammaproteobacteria (Oceanospiralles, Alteromonadales,Salinispaherales).e two largemoduleswere connected byOTUs classied asRhodobac-teriales which had negative correlations to one module and time lagged positive corre-lations to the other (Figure 4.4C). In the other largemodule the bacteriawere classied asphyla Flavobacteriia, Alphaproteobacteria (SAR 11 clade, Rhodobacterales, OCS116_clade),and Betaproteobacteria (Methylophilales). When the phylogenetic groups for the T4-likemyoviruseswere overlaid onto this network the phylogenetic patternswere no longergrouped into distinct modules (Figure 4.4D).In the association network containing the eukaryotic and marine picorna-like viralcommunities there were two large modules present (Figure 4.4E and F). In one groupthere were many marine picorna-like viral OTUs, and many eukaryotic nodes whichwere classied asAlveolates (Dinožagellata, Ciliophora, Protalveolata), Cryptomonadales(Hemiselmis, Teleaulax, FV18-2G7), Holozoa (Metazoa). Cryptophyceae (Kathablephar-idae) Picozoa (Picomonadea), Haptophyta (Prymnesiophyceae), Rhizaria (Cercozoa),Stramenopiles (Ochrophyta, Peronosporomycetes, MAST2, 4, 6, and 12). e secondmost highly connected module also had many marine picorna-like viral OTUs, and foreukaryotic nodes it contained nodes from the phyla Alveolates (Dinožagellata, Cilio-phora, Protalveolata), Holozoa (Metazoa), Picomonadea,Haptophyta (Prymnesiophyceae),Rhizaria (Cercozoa), Stramenopiles (Ochrophyta, Peronosporomycetes, MAST 12).emarine picorna-like viral OTUs 3 and 5 connected the modules.4.4.2 Temporal associations between communitiesNumber of edges between andwithin by season — e fall had the highest num-ber of shared edges even though the winter season had a higher number of nodes (Figure117chapter 44.5A).e summer had the lowest number of signicant edges by date and the lowestnumber of nodes. When comparing the number of edges between seasons, the fall andwinter had the greatest number of shared edges and the greatest number of nodes presentin both seasons. Summer to winter had the fewest number of edges shared even thoughthe summer to winter had more nodes than in the spring to summer.Figure 4.5: Interactions over time, between and within communities. A) Number ofpositive and negative edges using local similarity analysis (LSA). B) Number of edgeswithin and between seasons from local similarity networks. Edges by season divided bynumbers of samples by season.Number of edges between and within communities — ere were more connec-tions betweenOTUs fromdišerent communities thanwithin communities (Figure 4.5B).Most edges were positively correlated and this was consistent across subnetworks. esubnetworkwith the greatest number of edges within a single type of community was the118chapter 4T4-like community and between communities was the eukaryote tomarine picorna-likeviral network.Figure 4.6: Counts of triplets by environmental factor. Chlorophyll a, pH, and monthnumber were not displayed because of few to no connections. ere were no tripletsfound where one OTU was positively correlated to the environmental parameter andthe other OTU was negatively correlated.Edges counted by triplets by environmental factors — Temperature and phos-phate had the highest number of edges in all the communities (Figure 4.6). e T4-like myoviral to bacterial edges had no strong connections to bacterial abundance, viralabundance or dissolved oxygen and very few to nitrate+nitrite.e T4-like to bacterialedges had the highest number of edges associated to silicate concentration and all ofthese edges were negatively correlated.e marine picorna-like viral community to eu-karyotic community edges had few triplets linked to nitrate+nitrite and viral abundanceand they had the highest number of edges strongly linked to dissolved oxygen, silicateand temperature.e eukaryotic to bacterial edges had no strong connections to silicate119chapter 4and few strong links to dissolved oxygen, but in all other categories these communitiesgenerally had the most edges.4.4.3 Network statisticsTable 4.1: Network statistics. Table counting associations with Q value < 0.05, p value< 0.01, abs(LS) > 0.5. (continued below)OverallEukaryoticnetworkBacterialnetworkMarinepicorna-likevirusnetworkT4-likemyovirusnetworknodes 789 171 159 187 223edges 14918 892 846 1655 1861density 0.024 0.031 0.034 0.048 0.038modularity 0.37 0.42 0.41 0.47 0.42average degree 38 10 11 18 17max degree 172 41 41 55 70positive edges (%) 90 93 89 96 98negative edges (%) 10 7 11 4 2no delay (%) 49 76 92 90 76delay pos (%) 23 10 4 6 9delay neg (%) 29 14 3 4 15120chapter 4Eukaryote tobacterianetworkEukaryote to marinepicorna-like virusnetworkBacteria toT4-likemyovirusnetworknodes 285 308 274edges 1746 2804 886density 0.022 0.03 0.012modularity 0.22 0.3 0.39average degree 12 18 6max degree 59 77 34positive edges (%) 87 92 95negative edges (%) 13 8 5no delay (%) 55 38 13delay pos (%) 17 29 44delay neg (%) 29 33 42Of all the pairwise associations 3.77% were signicant and passed the false discovery test(Q value) (14918 out of 395605 potential associations). e strongest correlations werefound within the single community networks which were mostly composed of edgeswith no delay (between ~75-95% with no delay). is contrasted what was found inthe overall network and in the networks composed of two dišerent communities whichranged from 14% to 39% of edges with no delay (Table 4.1).When considering only the eukaryotic network there was higher network density(0.03), and modularity (0.42) than for the overall graph, but similar percentages of posi-tive (93%) and negative (7%) edges (Table 4.1).e marine picorna-like viral communi-ties had the highest density (0.05) and the gp23 communities had the second highestdensity (0.04). e viral community networks had a higher average degree (18 and121chapter 417) than the eukaryotic and bacterial communities (10 and 11) and a higher percent ofpositive edges (96% and 98%).4.4.4 Network statistics over timeNetworks among twocommunities — In the overall network, the node count withsignicant edges žuctuated between 50 to almost 300 nodes (Figure 4.8A and see alsoAppendix B). Conversely, all the subnetworks between communities had more stablenodes and edges over time. ere were žuctuations in density over time with the falland winter being the least dense. However, the bacterial to T4-like myovirus networkhad a spike in density in late March, and all communities had a spike in density in lateJune (which is related to the decrease in nodes, edges and diameter at this time). ebacterial to T4-like myovirus network had the highest modularity over time and theeukaryotic to bacterial had the lowest. For most of the communities the median degreeover time was stable, however, for the eukaryotic to bacterial network there was a spikein August, but there was no spike in nodes or edges.Networks among single communities — For the single community networks thenodes and edges were stable, except for the T4-like myoviruses which had more žuc-tuations in both edges and nodes over time (Figure 4.8B). All communities generallyshowed constant diameters up until late June, when there was a large decrease in di-ameters. e T4-like myovirus community generally had the highest diameter and themarine picorna-like community had the lowest.e marine picorna-like virus networkhad the highest density over time. Examining themedian degree, the eukaryotic networkhad the highest value in the summer and then it dropped oš in the fall. e T4-likemyoviruses were the most highly connected for most of the time series.122chapter 4Figure 4.7: Degree histograms. A) degree overall, B) degree between eukaryotes andmarine picorna-like viruses, C) degree between bacteria and T4-like myoviruses, D)degree between eukaryotes and bacteria,123chapter 4Figure 4.8: Network statistics over time of the subnetworks composed of two dišerentcommunities andof subnetworks within single communities. A) Two communitynetworks B) Single community networks. A and B include: Node counts, Edge counts,Diameter (note for all the combined networks this was 1), Density, Modularity, Averagepath length (note for all the combined networks this was 1), and Median degree. egreen vertical line corresponds to the spring bloom and grey lines correspond todivisions between seasons.124chapter 4Figure 4.9: Variation partitioning based on partial RDA. A) Eukaryotes grouped bytaxonomic order, B) Bacteria grouped by taxonomic order, C) Marine picorna-likeviruses grouped by phylogenetic clades from Chapter 3, D) T4-like myoviruses groupedby phylogenetic clades from Chapter 3. Parameters were divided into chemical, bioticand temporal parameters before being forward selected and then used in variationpartitioning.125chapter 44.4.5 Variation partitioningTo determine whether biological, chemical or temporal factors (or a combination ofthese) played a role in structuring the communities, partial redundancy variation par-titioning was performed (Figure 4.9). Both the eukaryotic and bacterial communitieswere not signicantly explained by models at the OTU-level, but had signicant mod-els at the order and family levels. In the eukaryotic community chemical parametersrepresented the largest amount of variation in the community (16%), followed by biotic(6%) and temporal factors (3%) (61% unexplained). For the bacterial communities, bio-logical factors explained the most variation (29%), followed by chemical (13%), and thentemporal (4%, unexplained 55%).e viral communities were not signicantly explained at the OTU-level by the en-vironmental parameters, but were signicant when the viral community relative abun-dance was summed using the phylogenetic groups from Chapter 3. For the marinepicorna-like viral communities, biological parameters explained the most of the varia-tion (23%), followed by temporal (13%) and then chemical (10%). Twenty percent of thevariation was overlapping between biological and chemical factors (residual 30%). Forthe T4-like myoviral communities less of the variation was explained, but the chemicalparameters explained the most (8%), followed by temporal (5%), then biological factors(1%), with 10% of overlap (residual 64%).4.5 discussionAmplicon sequencing of eukaryotic, bacterial and viral communities revealed signi-cant relationships in the temporal patterns of co-occurrence. Further analysis of theserelationships using network analysis, and the overall communities using redundancyanalysis and variation partitioning showed distinct communities (Figure 4.2, Figure 3and Figure 4.4), that were highly dynamic over time (Figure 4.8).126chapter 4e modules, which are highly connected communities of OTUs, detected similargroups in the networks as were determined phylogenetically (Figure 4.4). is furtherstrengthens the argument proposed in Chapter 3 that there is phylogenetic structuringto these communities. Also, the viral networks were more densely connected than thosefor bacteria and eukaryotes (Table 4.1 and Figure 4.8) suggesting a dišerent structurein these communities as was also seen in Chow et al. (2014). e overall network hadthe greatest number of connections per timepoint in the fall. Also when examiningstrongly correlated triplets of OTUs correlated with environmental factors, the bacterial-eukaryotic networks had more associations than the host-viral networks. One of theviral groups, the marine picorna-like viruses, had a large proportion of its varianceattributable to chemical factors, more than any of the other communities. is wassurprising, but similar results were seen in T4-like myoviral communities in coastalwaters (Chow et al., 2014). Overall, temporal partitioning of the environment was animportant driver of these communities.4.5.1 Environmental inžuence on individual OTUs compared to communitiesere was evidence for environmental control of the overall variance of the compositionand structure of planktonic communities. e inžuence of abiotic and biotic factorson the communities was examined since these are important factors for structuringbacterial, eukaryotic and viral communities (e.g. Cullen, 1991; Kirchman et al., 1991;Morris et al., 2005; Fuhrman et al., 2006; Fuhrman, 2009; Gilbert et al., 2009; Chowet al., 2014). Environmental inžuence at the community level dišered from the pairwiseLSA associations for the individual OTUs. Phosphate, nitrate+nitrite, temperature andsalinity had similarly large inžuences on the communities at the OTU- and community-levels. Nutrients, day length, temperature, salinity and chl a were important in all theviral, bacterial and eukaryotic communities, but not necessarily for the dynamics of theindividual OTUs. is is similar to Gilbert et al. (2011) who found that change in daylength could explain most of the variability in bacterial community diversity, therefore,127chapter 4seasonal change was deemed more important than trophic interactions. Chow et al.(2013) found that the bacterial community variance was related to salinity and chl a;protistan community variance was related to day length and bacterial abundance; andT4-like myoviral communities were related to day length, change in day length, salinityand temperature. ese examples and our results are in opposition to Lima-Mendezet al. (2015) who, based on their global spatial microbial networks, argued that bioticfactors are more important than abiotic for structuring communities since a minority ofthe associations they found could be explained by environmental factors.In themarine picorna-like viruses to eukaryotes environmental triplets, the dissolvedoxygen had the greatest number of edges andmostwere negatively correlated.is couldbe because of a specic type of OTU associated with a phytoplankton bloom since thedecrease in dissolved oxygen could show the demise of a phytoplankton bloom.Relationship of niche to environmental drivers of diversity — e nicheešects of OTUs correlated to the environment were examined as environmental triplets(Lima-Mendez et al., 2015). e OTUs in these triplets revealed that the bacterial andeukaryotic communities were more frequently strongly linked to each other and to theenvironment than were the host-virus-environment links. is points to a strongerrelationship of putative hosts to the abiotic factors, and that even at the OTU-level somebacteria and eukaryotes could be driven by similar parameters (Figure 4.6).e bacterialto T4-like myoviral communities had their strongest links with viral abundance. isis not surprising as the majority of viruses counted by žow cytometry would be bacte-riophage (Brussaard, 2004b). Phosphate and nitrate+nitrite had the greatest number oflinks in the triplet networks. Dišerences between community- andOTU-level dynamicswere likely from the dynamics of ephemeral OTUs. Changes in viral abundance and theviral community could have a partitioned ešect on the bacterial community e.g. thesechanges might not ašect the entire community in the same way (Chow et al., 2013).Within single networks, the links and modules were stronger than between dišerenttypes of amplicons. Gilbert et al. (2011) found stronger connections within bacterial128chapter 4networks than between bacteria and eukaryotes. However, they used dišerent resolu-tion data to examine the bacteria and the eukaryotes which could explain the dišerentpatterns. Similarly, Lima-Mendez et al. (2015) found that biotic factors were more ableto predict the composition of communities than abiotic.is hypothesis was examinedby looking at connections between communities and found that the highest number ofconnections were between eukaryotic and bacterial communities (Figure 4.5).e viralnetworks were more highly connected to themselves than the bacterial and eukaryoticcommunities (Table 4.1 and Figure 4.5).is would indicate that it is not always the sameviruses that are infecting the same hosts, or at least they are not completely dominantsince this analysis requiresOTUs to be presentmore than 20%of the time. Lima-Mendezet al. (2015) looked at the connections of the viruses and found that 43% only interactwith one host OTU and the rest only interact with a few. In the marine picorna-likecommunities, 8% of OTUs were connected to one putative host OTU and 24% wereconnected to ve or less (Figure 4.7). In the T4-like myoviral to bacterial networks 22%of OTUs were connected to one putative host OTU and 58% were connected to ve orless. is is in contrast to the overall network where 4% were connected only to oneOTU and 18% to ve or less. Considering only the connections between the eukaryoticand bacterial communities, 10%were connected to only one OTU and 38% to ve or less.Chow et al. (2013) found that protist-bacteria networks were composed of many smallhubs whereas the virus-bacteria was one big network. Also using LSA, Chow et al. (2014)saw many T4-like myoviral OTUs that were correlated to bacterial OTUs and some T4-like myoviral OTUs correlated strongly to multiple hosts suggesting a broad host range.Together these observations are consistent with the broader host ranges observed for theT4-like viruses compared to the more host-specic picorna-like viruses.e overall amount of variation explained in the communities was low. Variancepartitioning found that the marine picorna-like viral communities were more explainedby chemical parameters than the eukaryotic communities; while this was surprising itis concordant with what was found in T4-like myoviral communities at SPOT (Chow129chapter 4et al., 2014)where the community variationwasmore explained by the environment thanthe bacterial or eukaryotic communities. e bacterial communities in our study weremore inžuenced by biotic parameters such as chlorophyll a and bacterial abundancethan the eukaryotes. e marine picorna-like viral community had the most distinctseasonal progression (see Chapter 3 Adonis tests), so it not surprising that it has thehighest variance attributed to time.Two large negatively correlated modules detected for putative host net-works — ere were two large negatively correlated modules in the bacterial, eukary-otic and bacterial-eukaryotic networks (Figure 4.2 and Figure 4.4). Cram et al. (2015)suggest that modules can show dišerences in community states and these modules canrepresent seasonal patterns. In the eukaryotic, bacterial and bacterial to eukaryoticnetworks there were no taxonomic patterns found in the modules (Figure 4.2), which issimilar to results in Williams et al. (2014). Conversely, modules detect the main phylo-genetic groups in the viral-only networks (Figure 4.3).e viral OTUs sequenced are asmaller subset of the total viral community than the eukaryotic and bacterialOTUs are oftheir respective communities, so there is a greater chance that they would be temporallylinked or are following specic distributions of hosts.4.5.2 Network statistics over time: temporal nichese overall network was determined to have a “small world topology” because of thedegree distribution (Figure 4.7) and short path lengths found throughout the network(Table 4.1). is type of topology is quite common in microbial networks (e.g. Steeleet al., 2011; Chow et al., 2014; Cram et al., 2015; Lima-Mendez et al., 2015; Comte et al.,2015).As with other microbial time series, e.g. Chow et al. (2013), the highest richness ofOTUs and of community similarity was seen from Nov-Feb (parts of fall and winterseasons).e fall was the most highly connected network (Figure 4.5). In the northern130chapter 4hemisphere bacterial diversity is highest in winter at higher latitudes (Ladau et al., 2013).Ladau et al. (2013) found that the high richness in these communities was predictedby closeness to the thermocline, phosphate concentration, and day length. Short pho-toperiods are associated with high richness (Gilbert et al., 2011; Ghiglione and Murray,2012). Tan et al. (2013) describes how temporal niches are important for bacteria andcan promote diversity in these communities. us dynamic communities allow thisdiversity to persist or accumulate in the fall/winter. One hypothesis is that with similarenvironmental parameters over the fall/winter there was more time for the organismsto diversify or more growth of organisms adapted to the environmental conditions. Innetworks, resistance and stability comes fromhaving small world properties (Peura et al.,2015).ese small world networks, like most of the microbial networks found, are stablebecause of these properties and this could explain why there are not dramatic shiŸs overtime even with highly dynamic environmental parameters. If one organism is lost, thenetwork and its connections are not dramatically changed.4.5.3 CaveatsAlthough the challenges with viral gene markers and PCR in general were previouslydiscussed in the Introduction (p. 22), there are specic caveats for this network analysisof the co-occurrence of OTUs and the variance in communities over time. Cautionmustbe taken in interpreting the associations discovered in co-occurrence studies since asso-ciation does not necessarily mean interaction (Schluter, 1984). Additionally, only thosetaxa that occurred more than 30% of time were examined; so although individual OTUswere examined in the network analysis, these were abundant and common membersof the community. is excluded ephemeral associations, which could have includedimportant host-virus associations. However, it was preferred to lter the dataset conser-vatively rather than to add many false positives.131chapter 44.5.4 ConclusionsA one-year time series at Jericho Pier, in Vancouver B.C., provided a dynamic settingwithin which to examine the potential relationships of microbes and viruses. Basedon this study of the viral, bacterial and eukaryotic communities, their co-occurrence,and co-variance over time, partitioning of the environment is an important driver ofmicrobial community diversity. Overall there were more co-occurrences between dišer-ent communities than between like communities, reinforcing the idea that associationsamong dišerent communities are important. Highly connected groups within viral com-munitiesmatched the phylogenetic groups found inChapter 3, further strengthening thehypothesis of phylogenetic structuring of these communities. Based on the analysis ofthe environmental triplets, the environment likely a large role in ltering the host-viruspairs that occur in an environment and that occur seasonally.132chapter 5Viral and heterotrophic protistan control of aphytoplankton bloom5.1 summaryPhytoplankton blooms are important ecological events in the coastal and open oceanthat drive large drawdowns of CO2, stimulate the remineralization of nutrients and pro-mote succession in microeukaryotic and bacterial communities. Algal blooms can becomposed of multiple species and summer algal blooms in the Salish Sea can oŸen bedominated by members of the genus Heterosigma (raphidophyte). Studies have exam-ined how Heterosigma blooms progress over time and the potential allelopathic interac-tions with other organisms, but not the viral, bacterial and microeukaryotic communi-ties present during these blooms. During a summer algal bloom that was dominated bythe genusHeterosigma, high-throughput amplicon sequencing was used to examine thedynamics of the viral, bacterial and eukaryotic communities every other day for 18 days.For context these samples were compared to a one-year time series from the same site.At the peak biomass of the summer algal bloom, the bloom was dominated at the peakof chlorophyll by a population of Heterosigma that was undetectable two days later, butreturned 10 days later to form a smaller bloom. A succession of phytoplankton occurredduring the summer algal bloom in which dinožagellates formed smaller sub-bloomsduring the 18 days that did not co-occur with Heterosigma suggesting an antagonisticrelationship withHeterosigma or a preference for dišerent environmental conditions orselective predation of these phytoplankton. e bacterial and eukaryotic communitiesmirrored each other at the class level suggesting that bacterial communities are closely133chapter 5associated with specic taxa of phytoplankton.e succession of phytoplankton bloomsstimulated diversity in all communities and this increased diversity could be a responseto disturbance in the communities. Probing deeper into the diversity revealed strain-level dynamics where prominent OTUs in the bloom were decomposed into sub-typesusing Shannon entropy decomposition (oligotyping).e succession of these oligotypeswas linked to viral selective pressure early in the bloom and to protistan predation laterin the bloom, thus illuminating the strain-specic succession of phytoplankton speciesduring blooms.5.2 introductionPhytoplankton are a functional group that performs up to half of the photosynthesis onEarth (Field et al., 1998). Many phytoplanktonic taxa, in addition to being persistentmembers of the marine ecosystem, can also rapidly increase in cell number and overallbiomass to forming “blooms.” Blooms tend to occur when there in an inžux of nutrientsvia mixing or terrestrial inputs (Behrenfeld and Boss, 2014) and can cause large draw-downs of CO2 and can have large ešects on biogeochemical cycles (Alkire et al., 2012).Blooms can boost the rate of assimilation of nutrients like nitrogen and phosphorus inthe ecosystem and eventually increased bacterial remineralization (Buchan et al., 2014).Phytoplankton blooms can comprise single or mixed species of diatoms, dinožagel-lates, raphidophytes, and cyanobacteria (Harrison et al., 1983; Domingues et al., 2005),and typically occur in the spring, summer, and fall under specic conditions. Some ofthese blooms can be toxic to sh and other organisms and are a concern for aquacultureand shellsh collection (Nakamura et al., 1998; Oda et al., 1998). One important bloom-forming example is the raphidophyte,Heterosigma akashiwo, which is a eukaryotic algathat is toxic to sh (Horner et al., 1997; Lewitus et al., 2012; Powers et al., 2012). etoxicity to sh is mediated by gill-clogging mucus produced either by the alga or thesh itself (Nakamura et al., 1998; Oda et al., 1998), or by alga-produced neurotoxins134chapter 5(Khan et al., 1997; Ono et al., 2000). Heterosigma akashiwo tends to bloomunder specicconditions when the water temperature reaches 15°C and oŸen a decrease in salinitybelow 15 psu (Taylor andHaigh, 1993). Heterosigma can form cysts that can stay dormantand eventually seed a bloom (Powers et al., 2012). Blooms of Heterosigma have beendocumented in many studies along the B.C. coast (e.g. Haigh et al., 1992; Taylor et al.,1994), but the communities of microbes and viruses associated with the progression ofthese blooms is unexplored.Phytoplankton, especially bloom formers, can heavily inžuence their environmentsand other associated organisms. For example, there are oŸen correlations between bac-terial production and concentration of chlorophyll a (Cole, 1982; CroŸ et al., 2005; Sheret al., 2011). ShiŸs in the phytoplankton community (i.e. the phytoplankton in a ge-ographic area) can be followed by shiŸs in the bacterial community (Pinhassi et al.,2004), including changes in specic lineages such as the Flavobacteria, Bacteroides andAlphaproteobacteria (Buchan et al., 2014). Specic bacterial populations (i.e. the bac-terial species occurring in the same area) associated with phytoplankton blooms arefrequently seen in 16S rRNA amplicon studies (Buchan et al., 2014) and also in metage-nomic samples, providing information on specialized transporters, and metabolic path-ways (Teeling et al., 2012). Bacteria can support a bloom by recycling nutrients, but theyalso compete with phytoplankton for nutrients (Buchan et al., 2014). When there is alarge inžux of carbon from phytoplankton, it has been estimated that half of the carbonis processed by heterotrophic bacteria while the rest enters the food chain or sinks outof the photic zone (Buchan et al., 2014). During blooms, the structure of the bacterialcommunities may remain even (Delmont et al., 2014), but oŸen there are shiŸs in thecommunities to lineages with increased carbon cycling (Landa et al., 2016) and e›cientdegradation of by-products of phytoplankton (e.g. dimethylsulfoniopropionate (DMSP),transparent exopolymer particles (TEP), etc.) and decaying cells (Buchan et al., 2014).ShiŸs in the communities and populations of eukaryotes can occur during phyto-plankton blooms. For example, in blooms of the cosmopolitan dinožagellate Alexan-135chapter 5drium minutum, genetic diversity within the population can develop quickly (Dia et al.,2014). Conversely, during a bloom of the colony-forming haptophyte, Phaeocystis, thenumber of rare eukaryotic taxa and zooplankton biomass increased (Monchy et al., 2012).Understanding the role of bottom-up controls (e.g. nutrients) vs. top-down controls(grazing, viral lysis) is important for understanding the conditions under which bloomsoccur and progress. Blooms can be initiated by inputs of nutrients such as nitrogen,phosphorus, iron and silicate (Buchan et al., 2014) and by light availability. Ratios ofthese nutrients can also direct the establishment of specic bloom species (Buchan et al.,2014). Blooms can be terminated by bottom-up forces such as nutrient limitation (Daleet al., 1999; Blain et al., 2004; Mahadevan et al., 2012; Chiswell et al., 2015), or by top-down forces such as grazing by protists or zooplankton (Rosetta and McManus, 2003)or viral lysis (Bratbak et al., 1993, 1996; Lawrence and Suttle, 2004; Brussaard, 2004a).High abundances of viruses have oŸen been associated with phytoplankton blooms(Bratbak et al., 1990; Matteson et al., 2012). For example, in the bloom forming coccol-ithophore, Emiliania huxleyi, viruses (Emiliania huxleyi viruses (EhV)) are oŸen asso-ciated with blooms (Wilson et al., 2002a,b). Conversely, photosynthesis system reduc-tion has been observed when a bloom is controlled by viruses (Kimmance et al., 2014).Blooms can terminate with one viral genotype dominating; whereas, in other bloomsthere can be multiple genotypes (Schroeder et al., 2003; Martinez Martinez et al., 2007;Sorensen et al., 2009; Higheld et al., 2014). However, it is unknown what inžuencesthese dišerent scenarios.Heterosigma akashiwo strains can be infected by DNA viruses (Nagasaki and Ya-maguchi, 1997) and RNA viruses (Tai et al., 2003). e susceptibility of Heterosigmaakashiwo to viruses is oŸen dependent on the stage of the bloom when the strain wasisolated. Tarutani et al. (2000) found that during Heterosigma akashiwo blooms therewere multiple host strains. Viral control of these strains was apparent since strains iso-lated later in the bloom were more resistant to viruses than those isolated early in thebloom.136chapter 5Blooms disturb ecosystems by drastically changing the biological or physical environ-ment. Disturbances can disrupt an ecosystem, but can also promote diversity, increasedrichness and release of resources (Connell, 1978; Holt, 2008). Microbial communitiescan respond to disturbance events by not changing (stability), by returning to the waytheywere before the event (resilience), by changing composition but remaining function-ally the same, or by becomingmarkedly dišerent aŸer the disturbance (Allison andMar-tiny, 2008). As seen in the network analysis in Chapter 4 (p. 123) and references within,microbial communities appear to shiŸ in composition, but diversity is maintained overtime, thus, these communities appear resilient to change during a disturbance.Two main hypotheses were investigated in a summer algal bloom. First, since thereis evidence for viral control of phytoplankton blooms in mesocosm studies (Wilsonet al., 2002b; Martinez Martinez et al., 2007; Larsen et al., 2008), it suggests that high-throughput sequencing could be used to detect viral control of a naturally occurringsummer algal bloom. ese blooms can be dominated in this region by Heterosigmaand viruses that infect Heterosigma have been isolated and detected in the coastal wa-ters of British Columbia. Additionally, based on the strain-level specicity of virusesinfecting phytoplankton, it was hypothesized that there will be strain-level progressionof the bloom-forming phytoplankton as they experience selective pressures from specicviruses. Second, since phytoplankton blooms can be considered to be disturbances, andmicrobial communities are theorized to have high resilience, it was hypothesized thatthe bacterial, eukaryotic, and viral communities will behave as disturbed communitiesduring the bloom, show higher richness following the bloom, and then return to theirinitial state.To examine these hypotheses a summer algal bloom was sampled every other dayduring its initiation, peak anddemise at JerichoPier inVancouver, B.C. andhigh through-put sequencing of marker genes was used to follow temporal changes in the taxonomiccomposition of T4-like myoviruses, marine picorna-like viruses, bacteria and eukary-otes, and the richness, evenness, and composition of these communities as described in137chapter 5Chapter 3 (p. 139), bacteria and eukaryotes. e richness, evenness, and compositionof all communities were examined over time in addition to a variety of environmentalparameters. e strain-level variation of the bloom-forming taxa was also examinedusing Shannon Entropy decomposition (oligotyping Eren et al., 2013). ese changeswere examined in the context of a variety of environmental parameters.5.3 material and methods5.3.1 Sample collectione samples were collected and processed largely as outlined in Chapter 3 with thefollowing modications: Samples were collected from Jericho Pier (49° 16’36.73N, 123°12’05.41W) in British Columbia, Canada. Jericho Pier (JP) is adjacent to the shoreline,in a well-mixed location with mixed semi-diurnal tides. In order to get representativewater samples and enough material for viral extraction sixty liters of water was pumpedfrom the 1-m depth every two days at the daytime high tide from 21 June 2011 to 5 July2011 (8 samples). e samples from the full-year time series June 2010 and July 2011were used for comparison. Salinity and temperature were measured using a YSI probe(Yellow Springs, Ohio, USA). For all samples, the water was pre-ltered through a 65µm Nitex mesh and ltered sequentially through 142-mm diameter, 1.2-µm nominalpore-size glass-ber (GC50 Advantec MFS, Dublin, CA., USA) and 0.22 µm pore-sizepolyvinyldine (Millipore, Bedford, MA, USA) lters. e ltrate, containing the viralsize fraction, was concentrated to ~500 mL (viral concentrate) using tangential žowultraltration with a 30kDa MW prep-scale Spiral Wound TFF-6 cartridge (Millipore)(Suttle et al., 1991).5.3.2 NutrientsPhosphate, silicate and nitrate+nitrite concentrations were determined in duplicate 15-mL seawater samples ltered through0.45 µmpore-sizeHAlters (Millipore) and stored138chapter 5at -20°Cuntil air-segmented continuous-žowanalysis on aAutoAnalyzer 3 (Bran+Luebbe,Norderstedt, Germany). Chlorophyll a (Chl a) was determined in triplicate by ltering100 mL of seawater onto 0.45 µm pore-size HA lters (Millipore), and storing the ltersin the dark at -20°C until acetone extraction and then analysed žuorometrically (Parsonset al., 1984).5.3.3 Enumeration of bacteria and virusesSamples for viral and bacterial abundances were taken at each sampling point by x-ing duplicate cryovials containing 980µL of sample with nal concentration of 0.5%glutaraldehyde (EM-grade), freezing in liquid nitrogen and storing at -80°C until pro-cessing. Flow cytometry samples were processed as in Brussaard (2004b). Briežy, viralsamples were diluted 1:10 to 1:10 000 in sterile 0.1 µm ltered 1X TE, stained with SYBRGreen I (Invitrogen, Waltham, MA, USA) at a nal concentration of 0.5 x 10-4 of com-mercial stock, heated for 10 minutes at 80° C and then cooled in the dark for 5 minutesbefore processing. Bacterial samples were diluted up to 1:1000 in sterile 0.1 µm ltered1xTE, stainedwith SYBRGreen I at a nal concentration of 0.5 x 10-4 of commercial stock,and incubated in the dark for 15 minutes before processing. All samples were processedon a FACScalibur (Becton-Dickinson, Franklin Lakes, New Jersey, USA) with viral andbacterial samples run for 1 min at a medium or high žow rate, respectively. Event rateswere kept between 100 to 1000 events per second and green žuorescence and side scatterdetectors were used. Data were processed and gated using Cell-Quest soŸware (Becton-Dickinson).5.3.4 Extraction of viral nucleic acidse viral concentrate was ltered twice through 0.22µm pore-size Durapore PVDF l-ters (Millipore) in a sterile Sterivex lter unit (Millipore). e ltrate, containing viral-sized particles, was pelleted by ultracentrifugation (Beckman-Coulter, Brea, California,USA) in a SW40 rotor at 108 000 g for 5 h at 12°C.e pellet was resuspended overnight139chapter 5in 100 µL of supernatant at 4°C. To digest free DNA, the pellets were incubated with 1UµL-1DNAsewith a nal concentration 5mMMgCl2 for 3 h at room temperature. Nucleicacids were extracted using a Qiamp Viral Minelute spin kit (Qiagen, Hilden, Germany)according to the manufacturer’s directions.5.3.5 PCR amplication of T4-like myoviral marker geneTo target the marine T4-like myoviral capsid protein gene (gp23), PCRs were set up asin Filée et al. (2005). Briežy, each reaction mixture (nal volume, 50 µL) consisted of 2µL template DNA, 1x (nal concentration) PCR bušer (Invitrogen, Carlsbad, California,USA), 1.5 mMMgCl2, 0.2 mM of each deoxynucleoside triphosphate (Bioline, London,UK), 40 pmol of MZIA1bis and 40pmol of MZIA6, and 1 U Platinum Taq DNA poly-merase (Invitrogen) and program conditions as in Table 3.1.5.3.6 PCR amplication of picorna-like virus marker geneHalf of each viral extract was used to synthesize cDNA. To remove DNA, the extractedviral pellets were digested with DNase 1 (amplication grade) (Invitrogen).e reactionwas terminated by adding 2.5 mM EDTA (nal concentration) and incubating for 10min at 65°C. Complementary DNA (cDNA) was generated using Superscript III reversetranscriptase (Invitrogen) with random hexamers (50 ng µL-1) as per the manufacturer.PCR was performed using primer set MPL-2 to target the RdRp of marine picorna-like viruses (Culley and Steward, 2007). Each reaction mixture (nal volume, 50 µL)consisted of 50 ng of cDNA, 1x (nal concentration) PCR bušer (Invitrogen), 2 mMMgCl2, 0.2 mM of each deoxynucleoside triphosphate (Bioline, London, UK), 1 µMof each primer, and 1 U Platinum Taq DNA polymerase. e reaction was run in aPCR Express thermocycler (Hybaid, Ashford, UK) with program conditions as in Table3.1. Products were run on a 0.5X TBE 1% low melt gel, excised and extracted usingZymoclean Gel DNA Recovery Kit (Zymo) as per the manufacturer and a nal elutionstep of 2x10 µL EB bušer (Qiagen).140chapter 55.3.7 Filtration and extraction of marine bacteria and eukaryotesOne liter of seawater was taken from the sixty liters and ltered through a 0.22µm pore-size Durapore PVDF 47 mm lter (Millipore) in a sterile Sterivex lter unit (Millipore).e lter was either stored at -20°C until extraction or immediately extracted as follows.Filter extraction was as in Short and Suttle (2003). Briežy, lters were aseptically cut andincubated with lysozyme (Sigma-Aldrich, St. Louis, MO, USA) at a nal concentrationof 1mg mL-1 for 2 h at 37°C. Sodium dodecyl sulfate was added at a nal concentrationof 0.1 % (w/v) and each lter was put through three freeze-thaw cycles. Proteinase K(Qiagen) was then added to a nal concentration of 100 µg mL-1 and incubated for 1 h at55°C. DNA was sequentially extracted using equal volumes of phenol:chloroform:IAA(25:24:1), and chloroform:IAA (24:1). DNA was precipitated by adding NaCl to a nalconcentration of 0.3M and by adding 2X the extract volume of ethanol. Samples wereincubated at -20°C for at least 1 h and then centrifuged for 1 h at 20 000 g at 4°C. Extractswere washed with 70 % ethanol and were resuspended in 50 µL EB bušer (Qiagen).5.3.8 PCR amplication of bacterial and eukaryotic ribosomal sequencesPCR targeting eukaryotes used primers Euk1209f and Uni1392r as in Diez et al. (2001).ese primers target positions 1423 to 1641 and includes the variable region V8. Each re-action mixture (nal volume, 50 µL) consisted of 2 µL template, 1x (nal concentration)PCR bušer (Invitrogen), 1.5 mMMgCl2, 0.2 mM of each deoxynucleoside triphosphate(Bioline, London, UK), 0.3 µMof each primer, and 2.5 UPlatinumTaqDNApolymerase.e reaction was run in a PCR Express thermocycler (Hybaid, Ashford, UK) with pro-gram conditions as in Table 3.1.PCR targeting bacteria used primers 341F (Baker et al., 2003) and 907R (Muyzeret al., 1995).ese primers target the v3 to v5 regions. PCRs were run with the followingconditions: each reaction mixture (nal volume, 50 µL) consisted of 2 µL template, 1x(nal concentration) PCR bušer (Invitrogen), 1.5 mMMgCl2, 0.2 mM of each deoxynu-cleoside triphosphate (Bioline, London, UK), 0.4 µM of each primer, and 1 U Platinum141chapter 5Taq DNA polymerase. e reaction was run in a PCR Express thermocycler (Hybaid,Ashford, UK) with program conditions as in Table 3.1.5.3.9 Sequencing library preparationConstruction — PCR products not requiring gel excision were puried aŸer PCRusing AMPure XP beads (Beckman Coulter) at a ratio of 1.2:1 beads:product. Cleanedproducts were resuspended in 30 µL EB bušer (Qiagen). All products were quantiedusing the Picogreen dsDNA (Invitrogen) assay using Lambda DNA (Invitrogen) as astandard. Sample concentrations were read using iQ5 (Bio-Rad, Hercules, CA, USA)and CFX96 Touch systems (Bio-Rad). Pooled libraries were constructed using one ofeach of the amplicons at a concentration so that their molarity would be similar and thetotal concentration of the pool was ~700-900 ng. Pooled amplicons were concentratedusing AMPure XP beads (Beckman Coulter) at a ratio of 1.2:1 beads:product. NxSeqDNA sample prep kit 2 (Lucigen, Middleton, WI, USA) was used as per manufacturer’sdirections with either NEXTFlex 48 barcodes (BioO, Austin, USA) , NEXTžex 96 HTbarcodes (BioO), or TruSeq adapters (IDT, Coralville, Iowa). Libraries were cleaned upusing AMPure XP beads (Beckman Coulter) at a ratio of 0.9:1 beads:library.Quantification and quality control of libraries — Libraries were checked forsmall fragments (primer dimers and/or adapter dimers) using a 2100 Bionanalyzer (Ag-ilent, Santa Clara, CA, USA) with the High Sensitivity DNA kit (Agilent). e concen-tration of libraries was quantied using Picogreen dsDNA assay as above.e librarieswere quantied and checked for ampliable adapters using the Library QuanticationDNA standards 1-6 (Kappa Biosystems, Wilmington, USA) with the SsoFast EvaGreenqPCR supermix (Bio-Rad) using 10 µL EvaGreen master mix, 3 µL of 0.5uM F primer, 3µL of 0.5 µmRprimer and 4 µL of 1:1000, 1:5000 and 1:10000 dilutions of the libraries intriplicate on iQ5 (Bio-Rad) andCFX96Touch qPCRmachines. Cycling parameters wereas follows: 95°C for 30s, 35 cycles of 95°C for 5s, 60°C for 30s, and the melt curve gener-142chapter 5ation from 65°C to 95°C in 0.5°C steps (10s/step). Quantication from both Picogreenand qPCR assays were used to determine nal pooling of all libraries before sequenc-ing. Libraries were sequenced using 2x250bp PE Miseq (Illumina, San Diego, USA)sequencing at Génome Québec Innovation Centre at the McGill University (Montreal,QC, Canada), and 2x300bp PE Miseq (Illumina) sequencing at UBC PharmaceuticalSciences Sequencing Centre (Vancouver, BC, Canada) and at UCLA’s Genoseq (LosAngeles, CA, USA).5.3.10 Initial sequence processingLibraries were either split by the sequencing centre using CASAVA (Illumina) or split bythe user using the Miseq Reporter soŸware (Illumina). Sequence quality was initiallyexamined using FastQC (Andrews, 2015). Contaminating sequencing adapters wereremoved using Trimmomatic version 0.32 (Bolger et al., 2014) and the quality of thesequencing library further examined using fastx_quality (Gordon, 2014). Libraries werefurther split into individual amplicons (i.e. 18S, 16S, gp23 and MPL) and then, if the ex-pected overlap of the paired-end reads was 40bp or more, the paired reads were mergedusing PEAR (Zhang et al., 2014). Sequences were then quality trimmed using Trimmo-matic with the default quality settings. Sequences were aligned to known sequences(Silva 119 database (Quast et al., 2013) for 16S and 18S rRNA genes) using align.seqs inmothur 1.33.3 (Schloss et al., 2009) and those not aligned were removed. Viral sequenceswere queried using BLAST against databases containing the genemarkers of interest andsequences with an e-value below 10-3 were kept.5.3.11 Chimera checking, OTU picking and read normalizatione 16S and 18S rRNA gene sequences were checked for chimeras using USEARCHversion 8.0.1517 (Edgar, 2010) with the Gold reference database. Unique, non-chimericsequences were clustered at 97% similarity. Taxonomy for the 16S and 18S rRNA genesequences was assigned usingmothur (Wang-type algorithm) and the taxonomy in Silva143chapter 5119 (Quast et al., 2013). Resolution of targets and database prevented the assignment oftaxonomy below the level of genus for most OTUs (notably Heterosigma could not beidentied to species level). For the viral targets sequences were chimera-checked usingUSEARCH denovo and reference (Edgar, 2010). Viral sequences were then translatedusing FragGeneScan 1.20 (Rho et al., 2010). Viral reads were clustered using USEARCH(Edgar, 2010) at 95% similarity for MPL, and 95% similarity for T4-like myoviruses. Op-erational taxonomic unit (OTU) tables for all targets were constructed using USEARCH(Edgar, 2010). Sequences were normalized for this project by date and by target usingvegan (Oksanen et al., 2015).5.3.12 OligotypingOligotypes were chosen from three eukaryotic OTUs (97% similarity) that dominatedthe microeukaryotic community during the bloom. Oligotyping partitions the OTUinto subdivisions based on variability at positions of high nucleotide variability, thus itcan reveal ner-scale dynamics within OTUs. One OTUwas taxonomically classied asa raphidophyte and the other two as dinožagellates. To normalize sequencing ešort, 10000 random eukaryotic reads were selected and classied. From these, the reads fromthe top OTU classied as a raphidophyte, and the two most abundant dinožagellateOTUs were used for Shannon Entropy decomposition aka oligotyping (Eren et al., 2013).e same procedure was performed for the dinožagellate OTUs. A similar procedurewas performedwith themarine picorna-like viral OTUs, where 2500 random reads werechosen.5.3.13 Community similarity and Mantel testsBray-Curtis distance matrices were constructed from the normalized OTU abundancetables. Mantel tests were performed by comparing the community distance matrices toeach other and to distance matrices of environmental parameters using vegan.144chapter 55.4 results5.4.1 Environmental parametersDuring the two week bloom period there were dramatic shiŸs in some environmentalparameters (Figure 5.1). e most striking were measurements of chlorophyll a whichwere already high at 46.5 µg/L and then increased to 168.7 µg/L (+- error 26.56 µg/L)on 23 June 2011.e chlorophyll ameasurements were cross-referenced with secchi diskmeasurements and with chlorophyll estimates from the whole Strait of Georgia wherelarge blooms had also been reported for these dates (Irvine and Crawford, 2012).5.4.2 High viral abundance followed high bacterial abundancee viral abundance reached its lowest measured abundance (1.65× 107 viruses per mL)at the chlorophyll peak on 23 June 2011 but increased to amaximum of 7.145×107 virusesper mL on 29 June 2011.is peak in viral abundance lagged behind a peak in bacterialabundance two days earlier of 5.135 × 106 cells per mL (Figure 5.1).Temperature and salinity were stable during this period and ranged from 14.8°C to18.6°C and 7.6 psu to 13 psu. Nutrients (silicate, phosphate and nitrate+nitrite) were alsomostly stable except for a spike in phosphate on 27 June 2011 to 1.02 µM (up from 0.04µM).ere was a similar spike seen in nitrate+nitrite concentration beginning on 29June 2011.5.4.3 Richness and evenness of communities during the bloome richness (Figure 5.3) and Pielou’s evenness (Figure 5.4) of viral, bacterial, and mi-croeukaryotic communities žuctuated during the bloom. e lowest richness was ob-served in the eukaryotic communities during the peak of the Heterosigma blooms (21June and 1 July 2011) and the highest was observed on 3 July, nine days aŸer the peakof the rst Heterosigma bloom. e T4-like myoviruses had the highest richness on 27June (2 days before the peak in viral abundance) and the lowest on 3 July 2011.145chapter 5Figure 5.1: Environmental parameters during summer algal bloom 21 June -5 July2011. Error bars represent standard error of the mean. A) temperature, B) salinity, C)viral abundance, D) bacterial abundance, E) chlorophyll a, inset is chlorophyll a withshorter y-axis, F) silicate, G) phosphate, H) nitrate+nitrite I) dissolved oxygen (percentsaturation), J) pH. Stars (*) represent missing data. Grey vertical line demarcates thebeginning of summer, 22 June 2011.146chapter 5Figure 5.2: Environmental parameters during 1 year time series at Jericho Pier. Errorbars represent standard error of the mean. A) temperature, B) salinity, C) viralabundance, D) bacterial abundance, E) chlorophyll a, inset is chlorophyll a with shortery-axis, F) silicate, G) phosphate, H) nitrate+nitrite I) dissolved oxygen, and J) pH. Stars(*) represent missing data. e green vertical line indicates the spring diatom bloom.Grey vertical lines indicate seasons. Brown background indicates time of summer algalbloom.147chapter 5Figure 5.3: Richness of amplicons during the summer algal bloom (leŸ panels) andoverall 13 month time series (right panels). Richness calculated as number of observedOTUs aŸer normalizing by sampling ešort. A) and B) T4-like myoviruses C) and D)Bacteria E) and F) Eukaryotes. Grey arrows on x-axis indicate sampling time points foreach amplicon. Grey vertical lines indicate season boundary and the green vertical lineindicates the spring diatom bloom. Brown background indicates time of summer algalbloom.148chapter 5Figure 5.4: Evenness of amplicons during the summer algal bloom (leŸ panels) andoverall 13 month time series (right panels). Evenness calculated as Pielou’s evenness(Pielou, 1966). A) and B) T4-like myoviruses C) and D) Bacteria E) and F) Eukaryotes.Grey arrows on x-axis indicate sampling time points. Grey vertical lines indicateseason boundary and the green vertical line indicates the spring diatom bloom. Brownbackground indicates time of summer algal bloom.149chapter 55.4.4 Top 20 most abundant OTUs during the bloome highest richness in the T4-like myoviral communities was observed on June 27when the viral community composition was changed relative to the previous day (Figure5.5A). e dominant OTUs from the one-year time series were in low abundance orundetectable during the bloom, and the bloom community was not made up of the top20 most abundant OTUs found during the rest of the year (Figure 5.5A and B).e relative abundances of the top 20 bacterial OTUs in the community was morestable during the bloom than observed for the T4-like myoviruses. At the peak of therstHeterosigma bloom on June 23, a Flavobacteriaceae OTU (bacterial OTU 6) becamea dominant member of the bacterial community. In contrast, an alphabacterial OTUin the family Rhodobacteraceae (bacterial OTU 2) was persistent throughout the timeseries, but became dominant aŸer July 1, and made up more than 35% of the communityon July 5 (Figure 5.5C and D).e eukaryotic community also had large temporal dynamics. Before the peak inchlorophyll a the community was dominated by Heterosigma (eukaryotic OTU 2) (Fig-ure 5.5E). AŸer chlorophyll peaked, the OTUs increased from 189 to 336 (Figure 5.3Eand F), with dišerent OTUs dominating the communities. On June 29 Heterosigma(eukaryotic OTU 2) dominated the community again before being replaced by a higherdiversity of eukaryotic OTUs.e marine picorna-like viral community, as seen in Chapter 3 (p. 68), is much lesseven than the other communities (Figure 5.5G andH). Ahigher percentage of themarinepicorna-like virus communities was comprised of the top 20 OTUs than the T4-likemyoviruses, bacteria and microeukaryotes. In the picorna-like viral community oneOTU dominated the community during the summer algal bloom (OTU 288).150chapter 5Figure 5.5: Top 20 most relatively abundant OTUs from each community during thesummer algal bloom (leŸ panels) and over the entire time series (right panels). A) andB) T4-like myoviruses, C) and D) Bacteria, E) and F) Eukaryotes, G) and H) Marinepicorna-like viruses. Each contour represents a dišerent OTU. Grey arrows on x-axisindicate sampling time points. Grey vertical lines indicate season boundary and thegreen vertical line indicates the spring diatom bloom. Brown background indicates timeof summer algal bloom.151chapter 5Figure 5.6: Relative abundance of bacterial (top panels) and eukaryotic (bottom panels)OTUs classied by phyla, during the summer algal bloom (leŸ panels) and annually(right panels). Grey arrows on x-axis indicate sampling date. Grey vertical lines indicateseason boundary and the green vertical line indicates the spring diatom bloom. Brownbackground indicates time of summer algal bloom. In the bacterial community phylarepresenting less than 1%of relative abundance are grouped together into ‘other’ category.Classications were done using theWang algorithm as implemented inmothur (Schlosset al., 2009) and using the Silva 119 database (Quast et al., 2013)152chapter 5Figure 5.7: Relative abundance of bacterial (top panels) and eukaryotic (bottom panels)OTUs classied by class, during the summer algal bloom (leŸ panels) and annually (rightpanels). Grey arrows on x-axis indicate sampling time points. Grey vertical lines indicateseason boundary and the green vertical line indicates the spring diatom bloom. Brownbackground indicates time of summer algal bloom. In the bacterial community classesrepresenting less than 1% of relative abundance are grouped together into the ‘other’category. Classicationswere done using theWang algorithm as implemented inmothur(Schloss et al., 2009) and using the Silva 119 database (Quast et al., 2013)153chapter 55.4.5 Changes inOTUs or groups in bacterial and eukaryotic communitiesmirrored eachotherere were marked shiŸs in community composition during the study. For bacterialclasses, the most dramatic shiŸs occurred in Gammaproteobacteria, which spiked inrelative abundance during the bloom on June 21 and July 1 (Figure 5.7). At the peakof the bloom, the proportion of Betaproteobacteria and Flavobacteria was also higher,along with the rise and fall of a group of Cytophagia between the peak of the bloom andJune 29. e eukaryotes were dominated by the SAR supergroup throughout the year(Figure 5.6) with Stramenopiles and Alveolates switching in relative dominance duringthe summer algal bloom (Figure 5.7).5.4.6 Individual eukaryotic OTUse dominant eukaryotic OTU during the bloom was a raphidophyte (Figure 8A) thatdominated between June 21 and June 23 and then again on July 1, while a ciliate (familyIntramacronucleata) comprised >40% of the relative abundance on June 27 and 29 (Fig-ure 5.8C). Dinožagellates were at times dominant, and during the summer algal bloomthree dišerent OTUs dominated at dišerent times (Figure 5.8G).5.4.7 Oligotypes of viruses, and bloom-forming microeukaryotesAt the beginning of the summer algal bloom (June 21) two Raphidophyte oligotypesfrom OTU2 were detected (Figure 5.10 A); however, on the second bloom of this Raphi-dophyte (July 1), this OTU was dominated by one oligotype (Figure 5.10 A). e di-nožagellate OTU (classied as family Dinophyceae and genus Gymnodiniphycidae) thatbloomed ve days aŸer the start of the Heterosigma bloom was primarily comprised ofone oligotype (Figure 5.10 C); whereas, before the bloom more oligotypes were present(Figure 5.10 D). A second dinožagellate OTU “bloomed” two days aŸer Heterosigma,154chapter 5Figure 5.8: Change in relative abundance of OTUs for four eukaryotic orders duringthe summer algal bloom (leŸ panels) and throughout the year (right panels). A) and B)Raphidophytes, C) and D) Ciliates, E) and F) Diatoms, G) and H) Dinožagellates. Greyarrows on x-axis indicate sampling dates. Grey vertical lines indicate season boundaryand the green vertical line indicates the spring diatom bloom. Brown backgroundindicates time of summer algal bloom. Classications were done using the Wangalgorithm as implemented in mothur (Schloss et al., 2009) and using the Silva 119database (Quast et al., 2013)155chapter 5Figure 5.9: Oligotypes of specic OTUs pf marine picorna-like viruses during thesummer algal bloom (leŸ panels) and annually (right panels). A) and B) OTU 288,C) and D) OTU 3. Legend letters indicate dišerent oligotypes, letters refer to dišerentoligotypes in the top and bottom panels are not the same. Grey arrows on x-axis indicatesampling timepoints. Grey vertical lines indicate season boundary and the green verticalline indicates the spring diatom bloom. Brown background indicates time of summeralgal bloom. Oligotyping as in Eren et al. (2013).156chapter 5Figure 5.10: Oligotypes of OTUs during the bloom (leŸ panels) and annually (rightpanel). A) and B) Raphidophytes, C) and D) Dinožagellate OTU 1, E) and F)Dinožagellate OTU 2. Legend letters indicate dišerent oligotypes detected, lettersbetween top and bottom panels are not the same. Grey arrows on x-axis indicatesampling timepoints. Grey vertical lines indicate season boundary and the green verticalline indicates the spring diatom bloom. Brown background indicates time of summeralgal bloom. Classications were done using the Wang algorithm as implemented inmothur (Schloss et al., 2009) and using the Silva 119 database (Quast et al., 2013) andoligotyping as in Eren et al. (2013).157chapter 5and consisted of two oligotypes that were relatively abundant during the bloom anddominated thereaŸer (Figure 5.10 E and F).5.5 discussionHigh-throughput sequencing of samples taken every other day during a summer algalbloom revealed dynamics in themicrobial communities, including a quick succession ofseveral distinct dinožagellate OTUs.e bacterial community mirrored the eukaryoticcommunity at the order level, whilemarine picorna-like viruses showed shiŸs associatedwith the demise of the Heterosigma-dominated peak of bloom.5.5.1 Viral dynamics putatively explain host successionViruses can ašect the termination of phytoplankton blooms. In the bloom formingcoccolithophore, Emiliania huxleyi, typically one genotype of large dsDNA viruses goeson to dominate the community as the bloom progresses (Martinez Martinez et al., 2007;Sorensen et al., 2009). Using severalHeterosigmaakashiwo strains and largeDNAviruses(HAVs), Tarutani et al. (2000) found dišerent patterns of resistance where a virus iso-lated at the end of the bloom would be most ešective at lysing strains occurring duringthe bloom, but less ešective at lysing host strains isolated aŸer the bloom. Before thebloom, resistance to viral infection was low inHeterosigma akashiwo strains, while aŸerthe bloom it was high. erefore, viruses may ašect the clonal composition of pop-ulations, as previously demonstrated in cyanophage and Synechococcus in chemostatexperiments (Marston et al., 2012). ese studies reinforce the hypothesis that thereis a large cellular tness cost to maintaining resistance to viruses. In bacteriophage,mutations in receptors either in structure, density, or access by virus show tness costs(Lenski, 1988).erefore before the bloom there are less resistant cells in the population.In the samples from the one-year time series at Jericho Pier, there were high relativeabundances ofHeterosigma in the fall, as well asmarine picorna-like virusOTU 1 (Figure158chapter 55.5). Based on Chapter 3 (p. 68) marine picorna-like virus OTU1 and OTU288 wereclosely related toHeterosigma akashiwoRNA virus (Tai et al., 2003). During the summeralgal bloom, the marine picorna-like viral OTU 288 žuctuated with the raphidophytes(Figure 5.5).5.5.2 Potential role of protists in termination of phytoplankton sub-bloomsere was a second Heterosigma bloom on July 1, but none of the viral OTUs were asso-ciated with it, suggesting that the second bloom was terminated by either dinožagellatecompetition (Figure 5.8D), ciliate grazing (Figure 5.8B), infection by dišerent viruses, ornutrient limitation.Dinožagellates include microzooplankton grazers (e.g. Ceratium sp.) and phyto-plankton which can be mixotrophic (e.g. Dinophysis sp.), and which are oŸen a back-ground species during blooms of other phytoplankton taxa (Anderson et al., 2008). Sim-ilarly, the summer algal bloom at Jericho Pier was primarilyHeterosigma, but there werealso large signals from dinožagellate OTUs (Figure 5.8G and H).An increase in the relative abundance of ciliates, which was dominated by an OTUthat was unique to the bloom and classied in the family Intramacronucleata, occurredtwo days aŸer the rise of the dinožagellate OTU 2. ese ciliates could be consumingboth Heterosigma and dinožagellates that were in high abundance at this time and fa-cilitated a dišerent dinožagellate to bloom two days later (Figure 5.8G and H). isagrees with eld studies where tintinnids (a type of ciliate) showed opposite abundancedynamics to Heterosigma (Verity and Stoecker, 1982; Verity, 1987).5.5.3 Changes at the strain-level over timee main eukaryotic taxonomic groups during the bloom were generally also found inthe one-year time series (Figure 5.5).erefore, these OTUs were not ephemeral species.e raphidophyte, dinožagellate and ciliate OTUs that showed dynamics during thebloom, had one or two main OTUs that žuctuated during the summer algal bloom.159chapter 5To examine the dynamics within OTUs that are more than 97% similar, ShannonEntropy decomposition (oligotyping from Eren et al., 2013) was used to allow sub-OTUresolution.e oligotypes of raphidophyte OTUs decreased in diversity at the beginningof the bloom, andwere quickly dominated by one oligotype (Figure 5.10).e oligotypesof the picorna-like viral OTU 288 (Figure 9), which was closely related to HaRNAV(shown in Chapter 3), followed similar dynamics as those of the raphidophytes through-out the time series, with peaks in abundance in August, September and February, butOTU 288 also had an oligotype that was not seen in the rest of the year, but dominatedduring the bloom. ese dynamics are consistent with OTU 288 infecting one of thedominant bloom taxa, and potentially ašecting the duration and composition of thebloom.e dinožagellate OTUs had their greatest relative abundance (0.13%) on June 25,the day aŸer the rstHeterosigma bloom.e most abundant dinožagellate OTU-1 wasclassied to the family level as SCM15C8 and appeared to be a time generalist, as it waspresent throughout the time series.e oligotypes from the second dinožagellate, OTU-2, showed high evenness the day aŸer the rst Heterosigma bloom, however, towardthe end of the bloom (and aŸer the second smaller bloom ofHeterosigma) one oligotypedominated.e dinožagellateOTU-1was abundant aŸer the secondHeterosigma bloom,but had amuch simpler population structure thanOTU-2 with only one dominant oligo-type observed during the bloom and in the overall time series.us, there was evidencefor progression of strain-level taxa inžuenced by dišerent ecological parameters duringthis bloom.5.5.4 Bacterial communities mirror shiŸs in eukaryotic communitiesDuring the bloom, the bacterial communities appeared to mirror the shiŸs in the eu-karyotic communities (Figure 5.7, Mantel test on overall time series: 0.56, n=24). Mem-bers of the heterotrophic lineages Flavobacteriia, Rhodobacteracea (such as roseobac-ters), Alphaproteobacteria and Gammaproteobacteria (such as the Alteromonadaceae)160chapter 5dominate the bacterial communities associated with phytoplankton blooms (Buchanet al., 2014). During the bloom, the Gammaproteobacteria increased with Heterosigma,while the Alphaproteobacteria, Actinobacteria and Flavobacteria decreased. Flavobacte-ria may have becomemost abundant aŸer the peak of theHeterosigma bloom since theycan convert highmolecular weight (HMW) products of phytoplankton to lowmolecularweight (LMW) products (Buchan et al., 2014).e bacterial communities had the highest evenness of all communities throughouttwo-week period examined (Figure 5.3).is could indicate functional stability, wherebydišerent bacteria are using various phytoplankton products and thus the community isnever completely dominated by one bacterialOTU. SimilarlyDelmont et al. (2014) foundeven, stable and distinct bacterial communities associated with Phaeocystis sp. blooms.us these bloomsmay support dišerent niches for various bacteria, leading to evennessin the overall bacterial community.5.5.5 Community ecologyIntense biological events may create unique niches over time that enable dišerent organ-isms to exist in the communities, thus allowing distinct successional patterns. On June27, four days aŸer the peak in chl a (June 23rd) there was a spike in bacterial abundance,and two days later a peak in viral abundance (highest seen in this study) (Figure 5.1,Figure 5.2). In all the communities, the richness was greatest at the end of the two-week period, aŸer the succession of phytoplankton blooms (Figure 5.3, Figure 5.4).issuggests that these communities may have responded to a disturbance event, in thiscase a phytoplankton bloom, by increasing in diversity as predicted by the intermediatedisturbance hypothesis (Petraitis et al., 1989; Reynolds et al., 1993;Huston andDeAngelis,1994).is contrasts previous studies in freshwater when during a Prymnesium parvum(haptophyte alga) bloom there were no changes in either eukaryotic or bacterial speciesrichness, but community similarity was dišerent between bloom and non-bloom times(Jones et al., 2013).161chapter 5Communities can respond in dišerent ways to disturbances: they can stay function-ally the same (but composition will be dišerent), they can return to the same com-position aŸer the disturbance event (resilience) or they can stay the same (resistance)(Allison and Martiny, 2008). A phytoplankton bloom, although a natural part of thecommunity, can be considered a pulse disturbance (i.e. short duration, as opposed to a“press” disturbance which is longer-scale). For example, aquatic bacterial communitiesaŸer amechanicalmixing event in a lake, returned to pre-disturbance composition sevendays aŸer the mixing event (Shade et al., 2012b).roughout the sampling, the evennessremained stable. Richness was low immediately following the disturbance, but then in-creased aŸerwards.ey hypothesize that the biomass produced during the bloom wasa source of nutrients and carbon for the bacterial communities, and provided distinctniches and food sources for many dišerent organisms.eoretically, intermediate levelsof disturbance would generate the highest levels of species richness (Petraitis et al., 1989).Moderate nutrient inputs can stimulate eukaryotic community diversity (Spatharis et al.,2007) thus explaining the viral, bacterial and eukaryotic diversity aŸer the bloom.ere-fore, blooms of eukaryotic phytoplankton disturb the eukaryotic, bacterial, and subsetsof the viral communities, provoking strain-level succession and stimulating an increasein richness in all communities following the bloom.is agrees with the Behrenfeld andBoss (2014) hypothesis that abiotic factors do not control bloom dynamics, but ratherthat biotic factors related to imbalances in predator-prey relationships could provoke abloom.5.6 conclusionsPhytoplankton blooms have been well studied, but much less is known about bacterialand viral communities associated with blooms.ere was succession of eukaryotic phy-toplankton during a summer algal bloom.ere were sub-blooms ofHeterosigma and di-nožagellates throughout the sampling period, and evidence of viral and grazer control of162chapter 5the summer algal bloom. Sub-strain progression within the Heterosigma bloom showedevidence for viral pressure on the genetic diversity ofHeterosigma strains.e bacterialcommunities associated with the bloom, maintained evenness throughout the bloomeven while mirroring shiŸs in the eukaryotic communities.e bacterial communitiesappeared taxonomically well-adapted to using the products of phytoplankton blooms.Overall this study showed that the natural disturbance of a bloom stimulated changes insuccession and diversity in the viral, bacterial and microeukaryotic communities, andprovided a glimpse into the complex dynamics within phytoplankton blooms.163chapter 6Conclusion and Future Directions6.1 summaryTo examine important ecological questions related to the diversity and community struc-ture ofmarine viral communities, two ecologically important groups of viruses and theirhost communities were placed in a temporal context by using amplicon sequencing of aone-year time series. In addition, host datasets were examined for the potential inžuenceof the viruses.e identity of the members of the communities are not needed for measures of di-versity and community structure, however, using high-throughput sequencing, the viralsequences were determined, thus, even though the viruses were mostly unknown, theywere analysed in a phylogenetic context which provided more information about theirrelatedness and about the phylogenetic structure of viral communities. e sub-OTUdynamics, dynamics of reads within the chosen OTU percent similarity, of planktonicOTUs were determined during a eukaryotic phytoplankton bloom at a coastal site andshowed successional shiŸs related to viral and heterotrophic protist hosts.Considering the key ndings of these studies, this dissertation has advanced the eldof viral ecology in the following ways:In Chapter 2 I examined the patchiness and continued production ofmarine picorna-like viruses using pyrosequencing of the RNA dependent RNA polymerase (RdRp) ofviruses in the order Picornavirales. ere was patchiness in the OTUs with low over-lap between spatially proximate samples and high turnover in the mixed layer of thewater column.ese results implied continuous infection and lysis of eukaryotic phyto-164chapter 6plankton by these viruses.is study also showed the potential and power of ampliconsequencing for addressing ecological questions related to temporal dynamics.In Chapter 3 I examined the phylogenetic structure from two viral communities overtime which revealed the phylogenetic structuring of these communities and the natureof ephemeral vs. persistent OTUs. e results implied that viruses inžuence the com-position of the host communities, and that viral community structure is dependent onlifestyle (i.e. host range and burst size).ese examinations provided renements to theseed bank theory (including insight into phylogenetic dynamics of viral communities).In Chapter 4 I constructed co-occurrence networks from the one-year time seriesof the eukaryotic, bacterial, T4-like myoviral and picorna-like communities. is anal-ysis revealed the important role of environmental parameters in determining the co-occurrence of viruses and hosts. It also showed that eukaryotic and bacterial OTUs weremore strongly correlated to environmental factors than the viral OTUs. Communitiessampled in the fall were more strongly correlated to each other than any other seasonand the fall samples shared the greatest number of links with the winter timepoints.isdemonstrated a time of stability for these communities. Based on the analysis of theenvironmental triplets, the environment plays a large role in ltering the host-virus pairsthat occur in an environment and occur seasonally.In Chapter 5 I looked at the dynamics of the eukaryotic, bacterial and viral com-munities during a eukaryotic phytoplankton bloom (peak of bloom composed mostlyof Heterosigma). is project examined the ešect of disturbances (ecological perturba-tions) on microbial communities. e observed succession of sub-OTUs (oligotypes)was linked to viral selective pressure early in the bloomand to protistan predation later inthe bloom. Also, by examining the ne-scale dynamics it was observed, that eukaryoticphytoplankton blooms could be terminated by multiple dišerent factors over a shorttime period.165chapter 6is concluding chapter will discuss these advances in the context of the elds ofviral and microbial ecology. Moreover, this chapter presents the next steps arising fromthis work to further advance these elds.6.2 additions to the “seed bank” theoryAs seen in Chapters 3, 4, and 5, the “seed bank” model (or simply “Bank” model) describ-ing community structure (Breitbart and Rohwer, 2005; Brum et al., 2015) is a useful wayto examine the work presented in this dissertation.emodel posits thatmostmembersof the community are rare and there is a shuœing of the rare members that then becomeabundant based on the environment or on available hosts. is dissertation showedthat there is a phylogenetic component to this shuœing in viral communities, meaningthat closely related viruses seem to become abundant at the same time. us, there isan order to the “Bank” that was not previously incorporated whereby phylogenetically-related viruses show similar shiŸs in dominance of the community. e nature of viralreplication, which is generally more error prone than host genome replication, couldexplain the structure where there is one relatively abundant virus and many relatedviruses since it could be the results of erroneous replication.us diversity is generatedthrough errors and the shiŸs over time were based on the original dominant virus.erelated viruses present that arise originally from the abundant dominant virus.Studies have examined how the overall phylogenetic diversity changes in bacterialcommunities (Horner-Devine and Bohannan, 2006; Amend et al., 2016), but phylogenyhas neither been used to look at the shiŸs in the community nor to examine the relat-edness of viral communities. In this dissertation I showed that the viral communitieshave large shiŸs in dominance over time and the data also suggest that this phylogeneticstructure maintains the resilience and stability of these communities.166chapter 66.3 the nature of ephemeral and persistent otusover timeis dissertation examined which OTUs were persistent and which were ephemeral.is was mostly examined in Chapter 3, but Chapter 2 and Chapter 5 also examinedwhich OTUs persisted. Most members of the community were ephemeral, and thesememberswere usually present at low relative abundances.erewas high turnover in thecommunities, especially in the viral communities. e OTUs in the viral communitiesthat were persistent over time were oŸen those that were the most relatively abundantover time. One hypothesis is that these viral taxa may be so abundant that their loss dueto decay is slower over time (Wilhelm et al., 1998) or it could be that they are beingconstantly produced (as suggested in Chapter 2) by continually infecting organisms,albeit those organisms might not always be at high abundance.6.4 effect of environmental parameters ondetermining the co-occurrence of viruses andhostsFocusing on the common, abundant OTUs (and setting aside the ephemeral OTUs),much can be learned about the overall ecosystem from the co-occurrence of these OTUs.Food-webs inmarine systems are oŸen very complex (Legendre and Rivkin, 2008;Weitzet al., 2015). Microbial communities can have deeply interdependent relationships, how-ever, these relationships can be hard to examine. InChapter 4, based on network analysisof co-occurrence associations, the connections between viruses and hosts were mainlydriven by nutrients, temperature, salinity, viral abundance, and bacterial abundance.is is complementary towhatwas observed using variation partitioning, where changesin dišerent communities were driven by environmental parameters and by time. As dis-cussed in the Introduction and in Chapter 4, positive co-occurrences can be attributedto host-virus pairs, mutualism, or to the same preferred niche. Viral predation couldbe represented by positive or negative links, positive links could represent symbiosis, or167chapter 6shared niches, and negative links could represent predation, opposite niches, competi-tive exclusion.6.5 effect of disturbances on microbial communitiesMicrobes and viruses are thought to respond very quickly to pulses of nutrients (Buchanet al., 2014), increases in temperature (Kendrick et al., 2014), and other environmentalchanges.us it would be expected that if there was a disturbance in the ecosystem theseorganisms would be quickly ašected. As described in the Introduction there are manydišerent types of disturbances that can ašect microbial communities.is dissertationfocused on the ešect of a summer algal bloom onmicrobial communities. Although thedominant phytoplankton in this bloom,Heterosigma, has been widely studied in the lab,and as part of compositional phytoplankton surveys, there have been few studies thathave also looked at how the bacterial, viral, and overall eukaryotic communities changeduring a summer algal bloom.6.5.1 Resilience and stabilityOver the course of the eukaryotic phytoplankton bloom the richness and evenness of thebacterial community were stable, but the communities shiŸed in composition duringthat time.is has been documented for other phytoplankton using community nger-printing and amplicon sequencing, but not yet for summer algal blooms dominated byHeterosigma. AŸer the bloom all communities showed higher richness than immediatelypreceding the bloom. If systems with higher richness are considered to be more produc-tive, there was an overall increase in biodiversity and productivity aŸer this disturbance.is is in agreement with the intermediate disturbance hypothesis (Connell, 1978).e eukaryotic community displayed interesting dynamics: rst aHeterosigmaOTUbloomed, then there were two separate sub-blooms of dinožagellates, and nally a sec-ond bloom of the Heterosigma OTU (same oligotype) six days later. ese sub-bloom168chapter 6dynamics have, as of yet, received little attention. It was observed that these disturbancescan ašect the eukaryotic community dišerently than the bacterial community. For somebacterial populations, the disturbance acted like an input of nutrients or carbon, andthus there were shiŸs in the overall community structure, but the community still sup-ported similar levels of richness. Whereas for the eukaryotic community it was a timeof competitive exclusion with high overall unevenness.6.6 implications• e phylogenetic relatedness plays an important role in the community assemblyof viral and microbial communities and thus should be considered when examin-ing community dynamics.• Co-occurrence of hosts and viruses can be driven by environmental parameterssuch as nutrients, salinity and temperature and thus specic niches are importantnot just for the occurrence of certain organisms, but for their interactions.• Phytoplankton blooms can be composed of many smaller sub-blooms of dišerentorganisms and these disturbances can generate diversity in all microbial and viralcommunities aŸer the bloom.• Overall microbial and viral diversity is driven by shiŸs in phylogenetically-relatedorganisms over time and dišerent environmental parameters.When examining species on a rank abundance curve it is important to investigatethe phylogeny to really understand how communities change over time and also couldgive insight into historical events of the community. Using only a one year time seriesit is hard to denitively identify hosts of viruses, but using high-throughput sequencingof longer data sets would enable deeper insights into these relationships. Nevertheless,this dissertation has shown putative host-virus interactions and the strong ešect of theenvironment on these interactions.169chapter 66.7 future workAs with many studies performed, there are new or improved technologies that open upnew or deeper avenues for the research. One of the relevant improvements for this studyis in the area of metagenomics. Metagenomics has been used in the marine setting (Ven-ter et al., 2004) and formarine viruses (Breitbart et al., 2002) since its inception, however,what has changed has been the amount of data retrieved from metagenomes and howmany metagenomes can be compared. Early metagenomes used Sanger sequencing toexamine genetic material, however, now with high-throughput sequencing, the numberof reads retrieved from any metagenome has greatly increased. With this increase, abetter representation of all of the communities and especially the rarer members can beachieved. Also, with the relative ease and low cost of processing metagenomes, it is nowpossible to analyse metagenomes of time series from viral and microbial communities.Additionally, there have been large improvements in the functional annotation ofmetagenomes (i.e. themetabolic or enzymatic pathways present) through tools likeMeta-cyc (Caspi et al., 2014) and Metapathways (Konwar et al., 2015). erefore, more eco-logical questions can be examined using the functional roles and diversity of the com-munities. is leads to trait-based approaches as dened in the recent review focus-ing on microbial traits (Martiny et al., 2015). Furthermore, applying the approaches inthis dissertation to a longer time series or reanalyzing previous data sets to examinethe temporal phylogenetic relatedness could provide deeper insights such as: 1) moreopportunities to conrm how typical the dynamics are that were observed during theHeterosigma bloom 2) the ability to more closely examine virus-host co-occurrencesenabling the identication of virus-host pairs and, 3) the ability to look at seasonalityof composition of the communities and which members are driving this seasonality.Finally, two relevant types of experiments with isolated viruses are prompted bythis dissertation. First, is the continued need and importance of viral isolation fromaquatic microbial hosts. e genetic diversity of both the T4-like myoviruses and themarine picorna-like viruses and lack of cultured representatives illustrated the impor-170chapter 6tance of isolating viral-host systems (or of using techniques in which viral sequence canbe specically associated to host sequence, such as in single-cell sequencing of bacteria).Genomes can be sequenced from these viral isolates, which would be complementaryapproach to amplicon and metagenomic approaches. Second, is the examination ofthe dynamics between viruses infecting the same host but from dišerent families orcontaining dišerent genetic material (RNA vs DNA). An easily imaginable examplewould be for Heterosigma akashiwo where there is a DNA virus (HAV (Nagasaki et al.,1994)) and a RNA virus (HaRNAV (Tai et al., 2003)) that infect it. Considering that theseviruses have dišerent lifestyles and properties, what would be the conditions for one tobemore successful than the other? Using two related DNA viruses Nissimov et al. (2016)determined that there was a “ght club” for the viruses, where one appeared to be better“competitor.” Apotential link illustrating that such dynamics could lead to co-infection isin the putative recombination of a ssRNAvirus and ssDNAvirus (Circovirus) discoveredinmetagenomes and then veried by long-range PCR from lakes inYellowstone (Diemerand Stedman, 2012). Examining the dynamics of DNA and RNA viruses of one hostcould provide a way to test the hypothesis of the r- vs. K-selected viruses (Suttle, 2007)and whether there are environmental or other conditions that lead one to be that oranother.6.8 conclusionis dissertation examined the dynamics of the bacteria, eukaryotes and subsets of theviruses at the coastal site of Jericho Pier in Vancouver, British Columbia, Canada us-ing high-throughput sequencing. By using an unprecedented combination of bacterial,eukaryotic and viral community data, this dissertation provided advances in the relat-edness of viruses over time, the drivers of host-virus relationships, the dynamics andrichness of coastal plankton communities during blooms, and updates to models of thecommunity structure of viral communities. Altogether, this dissertation has advanced171chapter 6knowledge about the community structure and dynamics of viral communities, and hasidentied important avenues to be explored in time series of microbes and viruses.172BibliographyAcinas S.G., Sarma-Rupavtarm R., Klepac-Ceraj V., and Polz M.F. 2005. PCR-inducedsequence artifacts and bias: insights from comparison of two 16s rRNA clone librariesconstructed from the same sample. Applied and Environmental Microbiology71:8966–8969.Adriaenssens E.M. and Cowan D.A. 2014. Using signature genes as tools to assessenvironmental viral ecology and diversity. Applied and Environmental Microbiology80:4470–4480.Alkire M.B., D’Asaro E., Lee C., Jane Perry M., Gray A., Cetinić I., Briggs N., RehmE., Kallin E., Kaiser J., and González-Posada A. 2012. Estimates of net communityproduction and export using high-resolution, Lagrangian measurements of O2, NO3-,and POC through the evolution of a spring diatom bloom in the North Atlantic. DeepSea Research Part I: Oceanographic Research Papers 64:157–174.Allen S. and Wolfe M. 2013. Hindcast of the timing of the spring phytoplankton bloomin the Strait of Georgia, 1968-2010. Progress in Oceanography 115:6–13.Allison S.D. and Martiny J.B. 2008. Resistance, resilience, and redundancy in microbialcommunities. Proceedings of the National Academy of Sciences 105:11512–11519.Altschul S.F., GishW.,MillerW.,Myers E.W., andLipmanD.J. 1990. Basic local alignmentsearch tool. Journal of Molecular Biology 215:403–410.Amend A.S., Martiny A.C., Allison S.D., Berlemont R., Goulden M.L., Lu Y., TresederK.K., Weihe C., and Martiny J.B.H. 2016. Microbial response to simulated globalchange is phylogenetically conserved and linked with functional potential.e ISMEJournal 10:109–118.AmendA.S., Seifert K.A., andBrunsT.D. 2010.Quantifyingmicrobial communitieswith454 pyrosequencing: does read abundance count? Molecular Ecology 19:5555–5565.Anderson C.R., Siegel D.A., Brzezinski M.A., and Guillocheau N. 2008. Controlson temporal patterns in phytoplankton community structure in the Santa BarbaraChannel, California. Journal of Geophysical Research 113.Andrews S. 2015. FastQC. http://www.bioinformatics.bbsrc.ac.uk/projects/fastqc/ Ver-sion 0.11.4.Angly F.E., Felts B., Breitbart M., Salamon P., Edwards R.A., Carlson C., Chan A.M.,Haynes M., Kelley S., and Liu H. 2006.e marine viromes of four oceanic regions.PLoS biology 4:e368.173bibliographyApprill A., McNally S., Parsons R., and Weber L. 2015. Minor revision to V4 regionSSU rRNA 806r gene primer greatly increases detection of SAR11 bacterioplankton.Aquatic Microbial Ecology 75:129–137.ArslanD., LegendreM., Seltzer V., Abergel C., and Claverie J.M. 2011. DistantMimivirusrelative with a larger genome highlights the fundamental features of Megaviridae.Proceedings of the National Academy of Sciences 108:17486–17491.Azam F., Fenchel T., Field J.G., Gray J.S., Meyer-Reil L.A., andingstad F. 1983.eecological role of water-column microbes in the sea. Marine ecology progress series.Oldendorf 10:257–263.Baker G., Smith J., and Cowan D. 2003. Review and re-analysis of domain-specic 16Sprimers. Journal of Microbiological Methods 55:541–555.Banks S.C., Cary G.J., Smith A.L., Davies I.D., Driscoll D.A., Gill A.M., LindenmayerD.B., andPeakall R. 2013.Howdoes ecological disturbance inžuence genetic diversity?Trends in Ecology & Evolution 28:670–679.Barberán A., Bates S.T., Casamayor E.O., and Fierer N. 2012. Using network analysisto explore co-occurrence patterns in soil microbial communities.e ISME journal6:343–351.Behrenfeld M.J. and Boss E.S. 2014. Resurrecting the ecological underpinnings of oceanplankton blooms. Annual Review of Marine Science 6:167–194.Bellas C.M. and Anesio A.M. 2013. High diversity and potential origins of T4-typebacteriophages on the surface of Arctic glaciers. Extremophiles 17:861–870.Bellec L., Clerissi C., Edern R., Foulon E., Simon N., Grimsley N., and Desdevises Y.2014. Cophylogenetic interactions between marine viruses and eukaryotic picophyto-plankton. BMC Evolutionary Biology 14:59.Benson D.A., Karsch-Mizrachi I., Lipman D.J., Ostell J., and Wheeler D.L. 2007.GenBank. Nucleic Acids Research 35:D21–D25.BentonM.J. 2009.e Red Queen and the Court Jester: species diversity and the role ofbiotic and abiotic factors through time. Science 323:728–732.Berger S.A., Krompass D., and Stamatakis A. 2011. Performance, accuracy, andweb server for evolutionary placement of short sequence reads under maximumlikelihood. Systematic Biology 60:291–302.Bergh O., Borsheim K.Y., Bratbak G., and Heldal M. 1989. High abundance of virusesfound in aquatic environments. Nature 340:467–468.BerryD. andWidder S. 2014. Decipheringmicrobial interactions and detecting keystonespecies with co-occurrence networks. Frontiers in Microbiology 5.Bettarel Y., Bouvier T., and Bouvy M. 2009. Viral persistence in water as evaluated froma tropical/temperate cross-incubation. Journal of Plankton Research 31:909–916.174bibliographyBlain S., Guieu C., Claustre H., Leblanc K., Moutin T., Quéguiner B., Ras J., and SarthouG. 2004. Availability of iron and major nutrients for phytoplankton in the northeastAtlantic Ocean. Limnology and Oceanography 49:2095–2104.Bokulich N.A., Subramanian S., Faith J.J., Gevers D., Gordon J.I., Knight R., Mills D.A.,and Caporaso J.G. 2013. Quality-ltering vastly improves diversity estimates fromIllumina amplicon sequencing. Nature Methods 10:57–59.BolgerA.M., LohseM., andUsadel B. 2014. Trimmomatic a žexible trimmer for Illuminasequence data. Bioinformatics 30:2114–2120.Borcard D., Legendre P., and Drapeau P. 1992. Partialling out the spatial component ofecological variation. Ecology 73:1045.Bouvier T. and del Giorgio P.A. 2007. Key role of selective viral-induced mortality indeterminingmarine bacterial community composition. EnvironmentalMicrobiology9:287–297.Bratbak G., Egge J., and Heldal M. 1993. Viral mortality of the marine alga Emilianiahuxleyi (Haptophyceae) and termination algal blooms. Marine Ecology ProgressSeries 93:39–48.Bratbak G., Heldal M., Norland S., and ingstad T.F. 1990. Viruses as partners inspring bloom microbial trophodynamics. Applied and Environmental Microbiology56:1400–1405.Bratbak G., Wilson W., and Heldal M. 1996. Viral control of Emiliania huxleyi blooms?Journal of Marine Systems 9:75–81.BreitbartM. 2012.Marine viruses: truth or dare. Annual Review ofMarine Science 4:425–448.Breitbart M. and Rohwer F. 2005. Here a virus, there a virus, everywhere the same virus?Trends in Microbiology 13:278–284.Breitbart M., Salamon P., Andresen B., Mahašy J.M., Segall A.M., Mead D., AzamF., and Rohwer F. 2002. Genomic analysis of uncultured marine viral communities.Proceedings of the National Academy of Sciences 99:14250–14255.Brown C.T., Hug L.A., omas B.C., Sharon I., Castelle C.J., Singh A., Wilkins M.J.,Wrighton K.C., Williams K.H., and Baneld J.F. 2015. Unusual biology across a groupcomprising more than 15% of domain Bacteria. Nature 523:208–211.Brum J.R., Ignacio-Espinoza J.C., Roux S., Doulcier G., Acinas S.G., Alberti A., ChašronS., Cruaud C., de Vargas C., Gasol J.M., and others. 2015. Patterns and ecologicaldrivers of ocean viral communities. Science 348:1261498.Brum J.R., Schenck R.O., and Sullivan M.B. 2013. Global morphological analysis ofmarine viruses shows minimal regional variation and dominance of non-tailedviruses.e ISME journal 7:1738–1751.175bibliographyBrum J.R. and Sullivan M.B. 2015. Rising to the challenge: accelerated pace of discoverytransforms marine virology. Nature Reviews Microbiology 13:147–159.Brussaard C.P. 2004a. Viral control of phytoplankton populations—a review. Journal ofEukaryotic Microbiology 51:125–138.Brussaard C.P.D. 2004b. Optimization of procedures for counting viruses by žowcytometry. Applied and Environmental Microbiology 70:1506–1513.Brussaard C.P.D., Noordeloos A.A.M., Sandaa R.A., Heldal M., and Bratbak G. 2004.Discovery of a dsRNA virus infecting the marine photosynthetic protistMicromonaspusilla. Virology 319:280–291.BuchanA., LeCleir G.R., Gulvik C.A., andGonzález J.M. 2014.Master recyclers: featuresand functions of bacteria associated with phytoplankton blooms. Nature ReviewsMicrobiology 12:686–698.Butina T.V., Belykh O.I., Potapov S.A., and Sorokovikova E.G. 2013. Diversity of themajor capsid genes (g23) of T4-like bacteriophages in the eutrophic Lake Kotokel inEast Siberia, Russia. Archives of Microbiology 195:513–520.Capella-Gutierrez S., Silla-Martinez J.M., and Gabaldon T. 2009. trimAl: a tool forautomated alignment trimming in large-scale phylogenetic analyses. Bioinformatics25:1972–1973.Caporaso J.G., Kuczynski J., Stombaugh J., Bittinger K., Bushman F.D., Costello E.K.,Fierer N., Pena A.G., Goodrich J.K., and Gordon J.I. 2010. QIIME allows analysis ofhigh-throughput community sequencing data. Nature Methods 7:335–336.Caporaso J.G., Lauber C.L., Walters W.A., Berg-Lyons D., Lozupone C.A., TurnbaughP.J., Fierer N., and Knight R. 2011. Global patterns of 16S rRNA diversity at a depth ofmillions of sequences per sample. Proceedings of the National Academy of Sciences108:4516–4522.Caspi R., Altman T., Billington R., Dreher K., Foerster H., Fulcher C.A., Holland T.A.,Keseler I.M., Kothari A., Kubo A., Krummenacker M., Latendresse M., Mueller L.A.,Ong Q., Paley S., Subhraveti P., Weaver D.S., Weerasinghe D., Zhang P., and KarpP.D. 2014.e MetaCyc database of metabolic pathways and enzymes and the BioCyccollection of Pathway/Genome Databases. Nucleic Acids Research 42:D459–D471.Chašron S., Rehrauer H., Pernthaler J., and von Mering C. 2010. A global network ofcoexisting microbes from environmental and whole-genome sequence data. GenomeResearch 20:947–959.Chen F. and Suttle C.A. 1995. Amplication of DNA polymerase gene fragments fromviruses infecting microalgae. Applied and Environmental Microbiology 61:1274–1278.176bibliographyChen F., Wang K., Huang S., Cai H., Zhao M., Jiao N., and Wommack K.E. 2009.Diverse and dynamic populations of cyanobacterial podoviruses in the ChesapeakeBay unveiled throughDNApolymerase gene sequences. EnvironmentalMicrobiology11:2884–2892.Chesson P. 2000. Mechanisms of maintenance of species diversity. Annual review ofEcology and Systematics 31:343–366.Chesson P. and Huntly N. 1997. e roles of harsh and žuctuating conditions in thedynamics of ecological communities.e American Naturalist 150:519–553.Chiswell S.M., Calil P.H.R., and Boyd P.W. 2015. Spring blooms and annual cycles ofphytoplankton: a unied perspective. Journal of Plankton Research 37:500–508.Chow C.E.T. and Fuhrman J.A. 2012. Seasonality and monthly dynamics of marinemyovirus communities: Marine myovirus community dynamics at SPOT. Environ-mental Microbiology 14:2171–2183.Chow C.E.T., Kim D.Y., Sachdeva R., Caron D.A., and Fuhrman J.A. 2014. Top-downcontrols on bacterial community structure: microbial network analysis of bacteria,T4-like viruses and protists.e ISME journal 8:816–829.Chow C.E.T., Sachdeva R., Cram J.A., Steele J.A., Needham D.M., Patel A., ParadaA.E., and Fuhrman J.A. 2013. Temporal variability and coherence of euphotic zonebacterial communities over a decade in the Southern California Bight. e ISMEJournal 7:2259–2273.Chow C.E.T. and Suttle C.A. 2015. Biogeography of Viruses in the Sea. Annual Reviewof Virology 2:41–66.Christen A. 2013. Shortwave Irradiance at Vancouver Sunset (Tower) 2011.http://ibis.geog.ubc.ca/~achristn/infrastructure/climate-data/STSDA2-01-2011-to-12-2011.html .Clasen J., Hanson C., Ibrahim Y., Weihe C., Marston M., and Martiny J. 2013. Diversityand temporal dynamics of Southern California coastal marine cyanophage isolates.Aquatic Microbial Ecology 69:17–31.Clasen J.L. and Suttle C.A. 2009. Identication of freshwater Phycodnaviridae and theirpotential phytoplankton hosts, using DNA pol sequence fragments and a genetic-distance analysis. Applied and Environmental Microbiology 75:991–997.Cole J.J. 1982. Interactions between bacteria and algae in aquatic ecosystems. AnnualReview of Ecology and Systematics 13:291–314.Cole J.R., WangQ., Fish J.A., Chai B., McGarrell D.M., Sun Y., BrownC.T., Porras-AlfaroA., Kuske C.R., and Tiedje J.M. 2014. Ribosomal Database Project: data and tools forhigh throughput rRNA analysis. Nucleic Acids Research 42:D633–D642.177bibliographyCombeM. and Sanjuán R. 2014. Variation in RNA virus mutation rates across host cells.PLoS Pathogens 10:e1003855.Comeau A.M., Arbiol C., and Krisch H.M. 2010. Gene network visualization andquantitative synteny analysis of more than 300 marine T4-like phage scašolds fromthe GOS metagenome. Molecular Biology and Evolution 27:1935–1944.Comte J., Lovejoy C., Crevecoeur S., and F. Vincent W. 2015. Co-occurrence patternsin aquatic bacterial communities across changing permafrost landscapes. Biogeo-sciences Discussions 12:10233–10269.Connell J.H. 1978. Diversity in tropical rain forests and coral reefs. Science 199:1302–1310.Cram J., Sun F., and Fuhrman J.A. 2013. Marine bacterial, archaeal, and protistanassociation networks. In Encyclopedia of Metagenomics, pages 1–10, Springer NewYork, New York, NY.Cram J.A., Xia L.C., Needham D.M., Sachdeva R., Sun F., and Fuhrman J.A. 2015.Cross-depth analysis of marine bacterial networks suggests downward propagationof temporal changes. ISME J 9:2573–2586.CroŸ M.T., Lawrence A.D., Raux-Deery E., Warren M.J., and Smith A.G. 2005. Algaeacquire vitamin B12 through a symbiotic relationship with bacteria. Nature 438:90–93.CrumpB.C., HopkinsonC.S., SoginM.L., andHobbie J.E. 2004.Microbial biogeographyalong an estuarine salinity gradient: combined inžuences of bacterial growth andresidence time. Applied and Environmental Microbiology 70:1494–1505.Csardi G. and Nepusz T. 2006. e igraph soŸware package for complex networkresearch. InterJournal Complex Systems:1695.Cullen J.J. 1991. Hypotheses to explain high-nutrient conditions in the open sea.Limnology and Oceanography 36:1578–1599.Culley A.I., Lang A., and Suttle C.A. 2003. High diversity of unknown picorna-likeviruses in the sea. Nature 424:1054–1057.Culley A.I., Lang A., and Suttle C.A. 2006. Metagenomic analysis of coastal RNA viruscommunities. Science 312:1795–1798.Culley A.I., Mueller J.A., BelcaidM.,Wood-Charlson E.M., Poisson G., and Steward G.F.2014.e characterization of RNAviruses in tropical seawater using targeted PCR andmetagenomics. mBio 5:e01210–14–e01210–14.Culley A.I. and Steward G.F. 2007. New genera of RNA viruses in subtropical seawater,inferred from polymerase gene sequences. Applied and Environmental Microbiology73:5937–5944.Dale T., Rey F., andHeimdal B.R. 1999. Seasonal development of phytoplankton at a highlatitude oceanic site. Sarsia 84:419–435.178bibliographyDarriba D., Taboada G.L., Doallo R., and Bangor D. 2011. ProtTest 3: fast selection ofbest-t models of protein evolution. Bioinformatics 27:1164–1165.Delmont T.O., Hammar K.M., DucklowH.W., Yager P.L., and Post A.F. 2014. Phaeocystisantarctica blooms strongly inžuence bacterial community structures in the Amund-sen Sea polynya. Frontiers in Microbiology 5.Derelle E., Ferraz C., Escande M.L., Eychenié S., Cooke R., Piganeau G., Desdevises Y.,Bellec L., Moreau H., and Grimsley N. 2008. Life-cycle and genome of OtV5, a largeDNA virus of the pelagic marine unicellular green algaOstreococcus tauri. PLoS ONE3:e2250.DeSantis T.Z., Hugenholtz P., Larsen N., Rojas M., Brodie E.L., Keller K., Huber T.,Dalevi D., Hu P., and Andersen G.L. 2006. Greengenes, a chimera-checked 16s rRNAgene database and workbench compatible with ARB. Applied and EnvironmentalMicrobiology 72:5069–5072.Dia A., Guillou L., Mauger S., Bigeard E., Marie D., Valero M., and Destombe C. 2014.Spatiotemporal changes in the genetic diversity of harmful algal blooms caused by thetoxic dinožagellate Alexandrium minutum. Molecular Ecology 23:549–560.Diemer G.S. and Stedman K.M. 2012. A novel virus genome discovered in an extremeenvironment suggests recombination between unrelated groups of RNA and DNAviruses. Biology Direct 7:13.Diez B., Pedros-Alio C., Marsh T.L., and Massana R. 2001. Application of denaturinggradient gel electrophoresis (DGGE) to study the diversity of marine picoeukaryoticassemblages and comparison of DGGEwith othermolecular techniques. Applied andEnvironmental Microbiology 67:2942–2951.Djikeng A., Kuzmickas R., Anderson N.G., and Spiro D.J. 2009. Metagenomic analysisof RNA viruses in a fresh water lake. PLoS One 4:e7264.Domingo E., Sheldon J., and Perales C. 2012. Viral quasispecies evolution. Microbiologyand Molecular Biology Reviews 76:159–216.Domingues R.B., Barbosa A., and Galvão H. 2005. Nutrients, light and phytoplanktonsuccession in a temperate estuary (the Guadiana, south-western Iberia). Estuarine,Coastal and Shelf Science 64:249–260.DuRandM.D., OlsonR.J., andChisholmS.W. 2001. Phytoplankton population dynamicsat the Bermuda Atlantic Time-series station in the Sargasso Sea. Deep Sea ResearchPart II: Topical Studies in Oceanography 48:1983–2003.Dutilh B.E., Cassman N., McNair K., Sanchez S.E., Silva G.G.Z., Boling L., Barr J.J.,Speth D.R., Seguritan V., Aziz R.K., Felts B., Dinsdale E.A., Mokili J.L., and EdwardsR.A. 2014. A highly abundant bacteriophage discovered in the unknown sequences ofhuman faecal metagenomes. Nature Communications 5.179bibliographyEdgar R.C. 2004. MUSCLE: multiple sequence alignment with high accuracy and highthroughput. Nucleic Acids Research 32:1792–1797.Edgar R.C. 2010. Search and clustering orders of magnitude faster than BLAST.Bioinformatics 26:2460–2461.Emerson J.B.,omas B.C., Andrade K., Heidelberg K.B., and Baneld J.F. 2013. Newapproaches indicate constant viral diversity despite shiŸs in assemblage structure in anAustralian hypersaline lake. Applied and Environmental Microbiology 79:6755–6764.Eren A.M., Maignien L., Sul W.J., Murphy L.G., Grim S.L., Morrison H.G., and SoginM.L. 2013. Oligotyping: dišerentiating between closely related microbial taxa using16S rRNA gene data. Methods in Ecology and Evolution 4:1111–1119.Faith D.P. 1992. Conservation evaluation and phylogenetic diversity. Biological Conser-vation 61:1–10.Faust K., Lahti L., Gonze D., de Vos W.M., and Raes J. 2015a. Metagenomics meetstime series analysis: unraveling microbial community dynamics. Current Opinionin Microbiology 25:56–66.Faust K., Lima-Mendez G., Lerat J.S., Sathirapongsasuti J.F., Knight R., HuttenhowerC., Lenaerts T., and Raes J. 2015b. Cross-biome comparison of microbial associationnetworks. Frontiers in Microbiology 6.Faust K. and Raes J. 2012. Microbial interactions: from networks to models. NatureReviews Microbiology 10:538–550.FieldC.B., BehrenfeldM.J., Randerson J.T., and Falkowski P. 1998. Primary production ofthe biosphere: integrating terrestrial and oceanic components. Science 281:237–240.Filée J., Tétart F., Suttle C.A., and Krisch H.M. 2005. Marine T4-type bacteriophages,a ubiquitous component of the dark matter of the biosphere. Proceedings of theNational Academy of Sciences 102:12471–12476.Frederickson C.M., Short S.M., and Suttle C.A. 2003.e physical environment ašectscyanophage communities in British Columbia inlets. Microbial Ecology 46:348–357.Fuhrman J.A. 1999. Marine viruses and their biogeochemical and ecological ešects.Nature 399:541–548.Fuhrman J.A. 2009. Microbial community structure and its functional implications.Nature 459:193–199.Fuhrman J.A., Cram J.A., and Needham D.M. 2015. Marine microbial communitydynamics and their ecological interpretation. Nature Reviews Microbiology 13:133–146.Fuhrman J.A., Hewson I., SchwalbachM.S., Steele J.A., BrownM.V., and Naeem S. 2006.Annually reoccurring bacterial communities are predictable from ocean conditions.Proceedings of the National Academy of Sciences 103:13104–13109.180bibliographyFujiki T., Matsumoto K., Mino Y., Sasaoka K., Wakita M., Kawakami H., Honda M.C.,Watanabe S., and Saino T. 2014. Seasonal cycle of phytoplankton community structureand photophysiological state in the western subarctic gyre of the North Pacic.Limnology and Oceanography 59:887–900.Gaston K.J. 2000. Global patterns in biodiversity. Nature 405:220–227.Ghiglione J.F. and Murray A.E. 2012. Pronounced summer to winter dišerences andhigher wintertime richness in coastal Antarctic marine bacterioplankton: Temporalvariation in Southern Ocean coastal bacterioplankton. Environmental Microbiology14:617–629.Gibbons S.M., Caporaso J.G., Pirrung M., Field D., Knight R., and Gilbert J.A.2013. Evidence for a persistent microbial seed bank throughout the global ocean.Proceedings of the National Academy of Sciences 110:4651–4655.Gilbert J.A., Field D., SwiŸ P., Newbold L., Oliver A., Smyth T., Somereld P.J., Huse S.,and Joint I. 2009.e seasonal structure of microbial communities in the WesternEnglish Channel. Environmental Microbiology 11:3132–3139.Gilbert J.A., Steele J.A., Caporaso J.G., Steinbrück L., Reeder J., Temperton B., Huse S.,McHardy A.C., Knight R., Joint I., Somereld P., Fuhrman J.A., and Field D. 2011.Dening seasonal marine microbial community dynamics.e ISME Journal 6:298–308.Giovannoni S., Temperton B., and Zhao Y. 2013. Giovannoni et al. reply. Nature 499:E4–E5.Gobet A., Böer S.I., Huse S.M., van Beusekom J.E., Quince C., Sogin M.L., BoetiusA., and Ramette A. 2012. Diversity and dynamics of rare and of resident bacterialpopulations in coastal sands.e ISME journal 6:542–553.Goldsmith D.B., Parsons R.J., Beyene D., Salamon P., and Breitbart M. 2015. Deepsequencing of the viral phoH gene reveals temporal variation, depth-specic compo-sition, and persistent dominance of the same viral phoH genes in the Sargasso Sea.PeerJ 3:e997.Gorbalenya A.E., Enjuanes L., Ziebuhr J., and Snijder E.J. 2006. Nidovirales: Evolvingthe largest RNA virus genome. Virus Research 117:17–37.Gordon A. 2014. fastx_toolkit. https://github.com/agordon/fastx_toolkit Version-0.0.14.Guillou L., Bachar D., Audic S., Bass D., Berney C., Bittner L., Boutte C., BurgaudG., de Vargas C., Decelle J., del Campo J., Dolan J.R., Dunthorn M., Edvardsen B.,Holzmann M., Kooistra W.H., Lara E., Le Bescot N., Logares R., Mahé F., MassanaR., Montresor M., Morard R., Not F., Pawlowski J., Probert I., Sauvadet A.L., Siano R.,Stoeck T., Vaulot D., Zimmermann P., and Christen R. 2013.e Protist RibosomalReference database (PR2): a catalog of unicellular eukaryote Small Sub-Unit rRNAsequences with curated taxonomy. Nucleic Acids Research 41:D597–D604.181bibliographyGustavsen J. 2015. RdRp 454 amplicons Jericho and SOG scripts.http://github.com/jooolia/RdRp_454_amplicons_Jericho_and_SOG Doi:10.5281/zenodo.12509.Gustavsen J. 2016. Temporal data scripts. https://github.com/jooolia/phylo_temporal_jerichoDOI Zenodo: TBA.Gustavsen J.A., Winget D.M., Tian X., and Suttle C.A. 2014. High temporal and spatialdiversity in marine RNA viruses implies that they have an important role in mortalityand structuring plankton communities. Frontiers in Microbiology 5.Haigh R., Taylor F., and Sutherland T. 1992. Phytoplankton ecology of Sechelt Inlet,a 9ord system on the British Columbia coast. I. General features of the nano- andmicroplankton. Marine Ecology Progress Series 89:117–134.Harrell F. 2014. Hmisc: A package of miscellaneous R functions. https://cran.r-project.org/web/packages/Hmisc/index.html Version 3.17-1.Harrison P.J., Fulton J.D., Taylor F.J.R., and Parsons T.R. 1983. Review of the biologicaloceanography of the Strait of Georgia: pelagic environment. Canadian Journal ofFisheries and Aquatic Sciences 40:1064–1094.Harvey E.L. and Menden-Deuer S. 2012. Predator-induced žeeing behaviors in phyto-plankton: a newmechanism for harmful algal bloom formation? PLoSONE 7:e46438.Harvey P. and Purvis A. 1991. Comparative methods for explaining adaptations. Nature351:619–624.HeldalM. and BratbakG. 1991. Production and decay of viruses in aquatic environments.Mar. Ecol. Prog. Ser 72:205–212.Hewson I., Steele J.A., Capone D.G., and Fuhrman J.A. 2006a. Temporal and spatialscales of variation in bacterioplankton assemblages of oligotrophic surface waters.Marine Ecology Progress Series 311:67–77.Hewson I., Vargo G.A., and Fuhrman J.A. 2003. Bacterial diversity in shallow olig-otrophicmarine benthos and overlying waters: ešects of virus infection, containment,and nutrient enrichment. Microbial Ecology 46:322–336.Hewson I., Winget D.M., Williamson K.E., Fuhrman J.A., and Wommack K.E. 2006b.Viral and bacterial assemblage covariance in oligotrophic waters of the West FloridaShelf (Gulf of Mexico). Journal of the Marine Biological Association of the UnitedKingdom 86:591–604.Higheld A., Evans C., Walne A., Miller P.I., and Schroeder D.C. 2014. How manyCoccolithovirus genotypes does it take to terminate an Emiliania huxleyi bloom?Virology 466-467:138–145.Hijmans R.J. 2015. geosphere: Spherical Trigonometry. http://CRAN.R-project.org/package=geosphere R package version 1.5-1.182bibliographyHolmes E. 2009. Evolution and emergence of RNA viruses. Oxford Series in Ecologyand Evolution, Oxford University Press, USA.Holmes E.C. 2010.e RNA Virus Quasispecies: Fact or Fiction? Journal of MolecularBiology 400:271–273.Holt R.D. 2008.eoretical perspectives on resource pulses. Ecology 89:671–681.Holt R.D. and Pickering J. 1985. Infectious disease and species coexistence: a model ofLotka-Volterra form.e American Naturalist 126:196–211.Horner R.A., Garrison D.L., and Plumley F.G. 1997. Harmful algal blooms and red tideproblems on the US west coast. Limnology and Oceanography 42:1076–1088.Horner-Devine M.C. and Bohannan B.J. 2006. Phylogenetic clustering and overdisper-sion in bacterial communities. Ecology 87:S100–S108.Huang S., Wilhelm S.W., Jiao N., and Chen F. 2010. Ubiquitous cyanobacterialpodoviruses in the global oceans unveiled through viral DNA polymerase genesequences.e ISME Journal 4:1243–1251.Hurwitz B.L. and Sullivan M.B. 2013.e Pacic Ocean Virome (POV): a marine viralmetagenomic dataset and associated protein clusters for quantitative viral ecology.PLoS ONE 8:e57355.Huse S.M., Dethlefsen L., Huber J., Welch D.M., Relman D., and Sogin M.L. 2008.Exploring microbial diversity and taxonomy using SSU rRNA hypervariable tagsequencing. PLoS Genetics 4:e1000255.Huston M.A. and DeAngelis D.L. 1994. Competition and coexistence: the ešects ofresource transport and supply rates.e American Naturalist 144:954–977.Ibarbalz F.M., Pérez M.V., Figuerola E.L.M., and Erijman L. 2014.e bias associatedwith amplicon sequencing does not ašect the quantitative assessment of bacterialcommunity dynamics. PLoS ONE 9:e99722.Irvine J. and Crawford W. 2012. State of physical, biological, and selected sheryresources of Pacic Canadian marine ecosystems in 2011. J.R. Irvine and W.R.Crawford.Jameson E., Mann N.H., Joint I., Sambles C., and Mühling M. 2011. e diversity ofcyanomyovirus populations along a North–South Atlantic Ocean transect.e ISMEjournal 5:1713–1721.Jia Z., Ishihara R., Nakajima Y., Asakawa S., and Kimura M. 2007. Molecular characteri-zation of T4-type bacteriophages in a rice eld. Environmental Microbiology 9:1091–1096.Jiang S.C. and Paul J.H. 1994. Seasonal and diel abundance of viruses and occurrenceof lysogeny/bacteriocinogeny in the marine environment. Marine Ecology-ProgressSeries 104:163–172.183bibliographyJohnson L.S., Eddy S.R., and Portugaly E. 2010. Hidden Markov model speed heuristicand iterative HMM search procedure. BMC bioinformatics 11:431.Jones A.C., Liao T.S.V., Najar F.Z., Roe B.A., Hambright K.D., and Caron D.A. 2013.Seasonality and disturbance: annual pattern and response of the bacterial andmicrobial eukaryotic assemblages in a freshwater ecosystem: Freshwater microbialcommunity structure. Environmental Microbiology 15:2557–2572.Kearse M., Moir R., Wilson A., Stones-Havas S., Cheung M., Sturrock S., BuxtonS., Cooper A., Markowitz S., Duran C., ierer T., Ashton B., Meintjes P., andDrummond A. 2012. Geneious Basic: An integrated and extendable desktop soŸwareplatform for the organization and analysis of sequence data. Bioinformatics 28:1647–1649.Kembel S.W., Cowan P.D., Helmus M.R., Cornwell W.K., Morlon H., Ackerly D.D.,Blomberg S.P., and Webb C.O. 2010. Picante: R tools for integrating phylogenies andecology. Bioinformatics 26:1463–1464.Kemp P.F. and Aller J.Y. 2004. Estimating prokaryotic diversity: when are 16S rDNAlibraries large enough? Limnology and Oceanography: Methods 2:114–125.Kendrick B.J., DiTullio G.R., Cyronak T.J., Fulton J.M., VanMooy B.A.S., and Bidle K.D.2014. Temperature-induced viral resistance in Emiliania huxleyi (Prymnesiophyceae).PLoS ONE 9:e112134.Khan S., Arakawa O., and Onoue Y. 1997. Neurotoxins in a toxic red tide ofHeterosigmaakashiwo (Raphidophyceae) in Kagoshima Bay, Japan. Aquaculture Research 28:9–14.Kimmance S., Allen M., Pagarete A., Martínez Martínez J., and Wilson W. 2014.Reduction in photosystem II e›ciency during a virus-controlled Emiliania huxleyibloom. Marine Ecology Progress Series 495:65–76.Kirchman D.L., Cottrell M.T., and Lovejoy C. 2010.e structure of bacterial commu-nities in the western Arctic Ocean as revealed by pyrosequencing of 16S rRNA genes:Arctic Bacteria in Winter and Summer. Environmental Microbiology 12:1132–1143.Kirchman D.L., Suzuki Y., Garside C., and Ducklow H.W. 1991. High turnover rates ofdissolved organic carbon during a spring phytoplankton bloom. Nature 352:612–614.Klopfer P. and MacArthur R. 1960. Niche size and faunal diversity. e AmericanNaturalist 94:293–300.Koeppel A.F. and Wu M. 2013a. Species matter: the role of competition in the assemblyof congeneric bacteria.e ISME journal 8:531–540.Koeppel A.F. and Wu M. 2013b. Surprisingly extensive mixed phylogenetic andecological signals among bacterial Operational Taxonomic Units. Nucleic AcidsResearch 41:5175–5188.184bibliographyKonwar K.M., Hanson N.W., Bhatia M.P., Kim D., Wu S.J., Hahn A.S., Morgan-LangC., Cheung H.K., and Hallam S.J. 2015. MetaPathways v2.5: quantitative functional,taxonomic and usability improvements. Bioinformatics 31:3345–3347.Koonin E.V. 1991. e phylogeny of RNA-dependent RNA polymerases of positive-strand RNA viruses. Journal of General Virology 72:2197–2206.Koonin E.V., Wolf Y.I., Nagasaki K., and Dolja V.V. 2008.e Big Bang of picorna-likevirus evolution antedates the radiation of eukaryotic supergroups. Nature ReviewsMicrobiology 6:925–939.Kunin V., Engelbrektson A., Ochman H., and Hugenholtz P. 2009. Wrinkles in the rarebiosphere: pyrosequencing errors can lead to articial inžation of diversity estimates.Environmental Microbiology 12:118–123.Kyle J.E., Eydal H.S., Ferris F.G., and Pedersen K. 2008. Viruses in granitic groundwaterfrom 69 to 450 m depth of the Äspö hard rock laboratory, Sweden.e ISME Journal2:571–574.Ladau J., Sharpton T.J., Finucane M.M., Jospin G., Kembel S.W., O’Dwyer J., KoeppelA.F., Green J.L., and Pollard K.S. 2013. Global marine bacterial diversity peaks at highlatitudes in winter.e ISME journal 7:1669–1677.LandaM., Blain S., Christaki U., Monchy S., and Obernosterer I. 2016. ShiŸs in bacterialcommunity composition associated with increased carbon cycling in a mosaic ofphytoplankton blooms. ISME J 10:39–50.Lang A.S., Culley A.I., and Suttle C.A. 2004. Genome sequence and characterization ofa virus (HaRNAV) related to picorna-like viruses that infects the marine toxic bloom-forming alga Heterosigma akashiwo. Virology 320:206–217.Lang A.S., Rise M.L., Culley A.I., and Steward G.F. 2009. RNA viruses in the sea. FEMSMicrobiology Reviews 33:295–323.Larsen A., Flaten G.A.F., Sandaa R.A., Castberg T.,yrhaug R., Erga S.R., Jacquet S.,and Bratbak G. 2004. Spring phytoplankton bloom dynamics in Norwegian coastalwaters: Microbial community succession anddiversity. Limnology andOceanography49:180–190.Larsen J.B., Larsen A., yrhaug R., Bratbak G., and Sandaa R.A. 2008. Response ofmarine viral populations to a nutrient induced phytoplankton bloom at dišerentpCO 2 levels. Biogeosciences 5:523–533.Larsson A. 2014. AliView: a fast and lightweight alignment viewer and editor for largedatasets. Bioinformatics 30:3276–3278.Lawrence J.E. and Suttle C.A. 2004. Ešect of viral infection on sinking rates ofHeterosigma akashiwo and its implications for bloom termination. Aquatic MicrobialEcology 37:1–7.185bibliographyLe Gall O., Christian P., Fauquet C.M., King A.M.Q., Knowles N.J., Nakashima N.,Stanway G., and Gorbalenya A.E. 2008. Picornavirales, a proposed order of positive-sense single-stranded RNA viruses with a pseudo-T = 3 virion architecture. Archivesof Virology 153:715–727.Lee C.K., Herbold C.W., Polson S.W., Wommack K.E., Williamson S.J., McDonaldI.R., and Cary S.C. 2012. Groundtruthing next-gen sequencing for microbialecology–biases and errors in community structure estimates from PCR ampliconpyrosequencing. PLoS ONE 7:e44224.Legendre L. and Rivkin R. 2008. Planktonic food webs: microbial hub approach.MarineEcology Progress Series 365:289–309.Legendre P. and Legendre L. 1998. Numerical Ecology.Lennon J.T., AanderudZ.T., Lehmkuhl B.K., and Schoolmaster JrD.R. 2012.Mapping theniche space of soil microorganisms using taxonomy and traits. Ecology 93:1867–1879.Lenski R.E. 1988. Dynamics of interactions between bacteria and virulent bacteriophage.In Advances in Microbial Ecology, pages 1–44, Springer.Levin S.A. 1992.e problem of pattern and scale in ecology: the Robert H. MacArthuraward lecture. Ecology 73:1943–1967.Lewitus A.J., Horner R.A., Caron D.A., Garcia-Mendoza E., Hickey B.M., Hunter M.,Huppert D.D., Kudela R.M., Langlois G.W., Largier J.L., Lessard E.J., RaLonde R.,Jack Rensel J., Strutton P.G., Trainer V.L., and Tweddle J.F. 2012. Harmful algal bloomsalong the North American west coast region: History, trends, causes, and impacts.Harmful Algae 19:133–159.Liang Y., Li L., Luo T., Zhang Y., Zhang R., and Jiao N. 2014. Horizontal and verticaldistribution of marine virioplankton: a basin scale investigation based on a globalcruise. PLoS ONE 9:e111634.Lima-MendezG., FaustK., HenryN., Decelle J., Colin S., Carcillo F., Chašron S., Ignacio-Espinosa J.C., Roux S., Vincent F., and others. 2015. Determinants of communitystructure in the global plankton interactome. Science 348:1262073.Logares R., Audic S., Bass D., Bittner L., Boutte C., Christen R., Claverie J.M., DecelleJ., Dolan J.R., Dunthorn M., Edvardsen B., Gobet A., Kooistra W.H., Mahe F., NotF., Ogata H., Pawlowski J., Pernice M.C., Romac S., Shalchian-Tabrizi K., Simon N.,StoeckT., Santini S., SianoR.,Wincker P., ZingoneA., Richards T.A., deVargasC., andMassana R. 2014. Patterns of rare and abundantmarinemicrobial eukaryotes. CurrentBiology 24:813–821.Loman N.J., Misra R.V., Dallman T.J., Constantinidou C., Gharbia S.E., Wain J., andPallen M.J. 2012. Performance comparison of benchtop high-throughput sequencingplatforms. Nature Biotechnology 30:434–439.186bibliographyLong J.R., Pittet V., Trost B., Yan Q., Vickers D., Haakensen M., and Kusalik A. 2014.Equivalent input produces dišerent output in the UniFrac signicance test. BMCbioinformatics 15:1.Lozupone C. and Knight R. 2005. Unifrac: a new phylogenetic method for comparingmicrobial communities. Applied and Environmental Microbiology 71:8228–8235.Lozupone C.A. and Knight R. 2007. Global patterns in bacterial diversity. Proceedingsof the National Academy of Sciences 104:11436–11440.Lozupone C.A. and Knight R. 2015. e unifrac signicance test is sensitive to treetopology. BMC Bioinformatics 16:1–3.López-Bueno A., Rastrojo A., Peiró R., Arenas M., and Alcamí A. 2015. Ecologicalconnectivity shapes quasispecies structure of RNA viruses in an Antarctic lake.Molecular Ecology 24:4812–4825.López-Bueno A., Tamames J., Velázquez D., Moya A., Quesada A., and Alcamí A. 2009.High diversity of the viral community from an Antarctic lake. Science 326:858–861.MacArthur R.H. and Wilson E.O. 1967. e theory of island biogeography, vol. 1.Princeton University Press.Mahadevan A., D’Asaro E., Lee C., and Perry M.J. 2012. Eddy-driven straticationinitiates north atlantic spring phytoplankton blooms. Science 337:54–58.Marchler-Bauer A., Lu S., Anderson J.B., Chitsaz F., Derbyshire M.K., DeWeese-ScottC., Fong J.H., Geer L.Y., Geer R.C., Gonzales N.R., Gwadz M., Hurwitz D.I., JacksonJ.D., Ke Z., Lanczycki C.J., Lu F., Marchler G.H., Mullokandov M., Omelchenko M.V.,Robertson C.L., Song J.S., anki N., Yamashita R.A., Zhang D., Zhang N., ZhengC., and Bryant S.H. 2010. CDD: a Conserved Domain Database for the functionalannotation of proteins. Nucleic Acids Research 39:D225–D229.Marston M.F., Pierciey F.J., Shepard A., Gearin G., Qi J., Yandava C., Schuster S.C.,Henn M.R., and Martiny J.B.H. 2012. Rapid diversication of coevolving marineSynechococcus and a virus. Proceedings of theNational Academy of Sciences 109:4544–4549.Marston M.F. and Sallee J.L. 2013. Genetic diversity and temporal variation in thecyanophage community infecting marine Synechococcus species in Rhode Island’scoastal waters. Applied and Environmental Microbiology 69:4639–4647.Marston M.F., Taylor S., Sme N., Parsons R.J., Noyes T.J.E., and Martiny J.B.H.2013. Marine cyanophages exhibit local and regional biogeography. EnvironmentalMicrobiology 15:1452–1463.Martinez Martinez J., Schroeder D.C., Larsen A., Bratbak G., and Wilson W.H. 2007.Molecular dynamics ofEmiliania huxleyi and cooccurring viruses during two separatemesocosm studies. Applied and Environmental Microbiology 73:554–562.187bibliographyMartiny J.B.H., Jones S.E., Lennon J.T., and Martiny A.C. 2015. Microbiomes in light oftraits: A phylogenetic perspective. Science 350:aac9323–aac9323.Massana R., Gobet A., Audic S., Bass D., Bittner L., Boutte C., Chambouvet A., ChristenR., Claverie J.M., Decelle J., Dolan J.R., Dunthorn M., Edvardsen B., Forn I., ForsterD., Guillou L., Jaillon O., Kooistra W.H.C.F., Logares R., Mahé F., Not F., Ogata H.,Pawlowski J., Pernice M.C., Probert I., Romac S., Richards T., Santini S., Shalchian-Tabrizi K., Siano R., Simon N., Stoeck T., Vaulot D., Zingone A., and de Vargas C.2015. Marine protist diversity in European coastal waters and sediments as revealedby high-throughput sequencing. Environmental Microbiology 17:4035–4049.Matsen F., KodnerR., andArmbrust E.V. 2010. pplacer: linear timemaximum-likelihoodand Bayesian phylogenetic placement of sequences onto a xed reference tree. BMCbioinformatics 11:538.MattesonA.R., Loar S.N., Pickmere S., DeBruyn J.M., EllwoodM.J., Boyd P.W., HutchinsD.A., and Wilhelm S.W. 2012. Production of viruses during a spring phytoplanktonbloom in the South Pacic Ocean near of New Zealand. FEMSMicrobiology Ecology79:709–719.Mayer J.A. and Taylor F.R. 1979. A virus which lyses the marine nanožagellateMicromonas pusilla. Nature 281:299–301.Middelboe M. 2000. Bacterial growth rate and marine virus–host dynamics. MicrobialEcology 40:114–124.Middelboe M., Hagström A., Blackburn N., Sinn B., Fischer U., Borch N., Pinhassi J.,Simu K., and Lorenz M. 2001. Ešects of bacteriophages on the population dynamicsof four strains of pelagic marine bacteria. Microbial Ecology 42:395–406.Middelboe M., Holmfeldt K., Riemann L., Nybroe O., and Haaber J. 2009. Bacterio-phages drive strain diversication in amarine Flavobacterium : implications for phageresistance and physiological properties. Environmental Microbiology 11:1971–1982.Miki T. and Jacquet S. 2010. Indirect interactions in the microbial world: specicitiesand similarities to plant–insect systems. Population Ecology 52:475–483.Miller M., Pfeišer W., Schwartz T., and others. 2010. Creating the CIPRES Science Gate-way for inference of large phylogenetic trees. In Gateway Computing EnvironmentsWorkshop (GCE), 2010, pages 1–8, IEEE.Milo R., Jorgensen P., Moran U., Weber G., and Springer M. 2010. BioNumbers thedatabase of key numbers in molecular and cell biology. Nucleic Acids Research38:D750–D753.Mizumoto H., Tomaru Y., Takao Y., Shirai Y., and Nagasaki K. 2006. Intraspecies hostspecicity of a single-stranded RNA virus infecting a marine photosynthetic protistis determined at the early steps of infection. Journal of Virology 81:1372–1378.188bibliographyMojica K.D. and Brussaard C.P. 2014. Factors ašecting virus dynamics and microbialhost-virus interactions inmarine environments. FEMSMicrobiology Ecology 89:495–515.Monchy S., Grattepanche J.D., Breton E., Meloni D., Sanciu G., Chabé M., Delhaes L.,Viscogliosi E., Sime-Ngando T., and Christaki U. 2012. Microplanktonic communitystructure in a coastal system relative to a Phaeocystis bloom inferred frommorpholog-ical and tag pyrosequencing methods. PLoS ONE 7:e39924.Moreau H., Piganeau G., Desdevises Y., Cooke R., Derelle E., and Grimsley N. 2010.Marine Prasinovirus genomes show low evolutionary divergence and acquisition ofprotein metabolism genes by horizontal gene transfer. Journal of Virology 84:12555–12563.Morris R.M., Vergin K.L., Cho J.C., Rappé M.S., Carlson C.A., and GiovannoniS.J. 2005. Temporal and spatial response of bacterioplankton lineages to annualconvective overturn at the Bermuda Atlantic Time-series Study site. Limnology andOceanography 50:1687–1696.Murray A.G. and Jackson G.A. 1992. Viral dynamics: a model of the ešects of size, shape,motion and abundance of single-celled planktonic organisms and other particles.Marine Ecology Progress Series 89:103–116.Muyzer G., Teske A., Wirsen C.O., and Jannasch H.W. 1995. Phylogenetic relationshipsof iomicrospira species and their identication in deep-sea hydrothermal ventsamples by denaturing gradient gel electrophoresis of 16S rDNA fragments. Archivesof Microbiology 164:165–172.MühlingM., Fuller N.J., Millard A., Somereld P.J., Marie D., WilsonW.H., Scanlan D.J.,Post A.F., Joint I., andMannN.H. 2005.Genetic diversity ofmarine Synechococcus andco-occurring cyanophage communities: evidence for viral control of phytoplankton.Environmental Microbiology 7:499–508.Nagasaki K., Ando M., Itakura S., Imai I., and Ishida Y. 1994. Viral mortality in thenal stage of Heterosigma akashiwo (Raphidophyceae) red tide. Journal of PlanktonResearch 16:1595–1599.Nagasaki K., Shirai Y., Takao Y., Mizumoto H., Nishida K., and Tomaru Y. 2005.Comparison of genome sequences of single-stranded RNA viruses infecting thebivalve-killing dinožagellate Heterocapsa circularisquama. Applied and Environmen-tal Microbiology 71:8888–8894.Nagasaki K., Tomaru Y., Katanozaka N., Shirai Y., Nishida K., Itakura S., andYamaguchi M. 2004. Isolation and characterization of a novel single-strandedRNA virus infecting the bloom-forming diatom Rhizosolenia setigera. Applied andEnvironmental Microbiology 70:704–711.189bibliographyNagasaki K. and YamaguchiM. 1997. Isolation of a virus infectious to the harmful bloomcausing microalga Heterosigma akashiwo (Raphidophyceae). Aquatic Microbial Ecol-ogy 13:135–140.Nakamura A., Okamoto T., Komatsu N., Ooka S., Oda T., Ishimatsu A., andMuramatsuT. 1998. Fish mucus stimulates the generation of superoxide anion by Chattonellamarina and Heterosigma akashiwo. Fisheries Science 64:866–869.Needham D.M., Chow C.E.T., Cram J.A., Sachdeva R., Parada A., and Fuhrman J.A.2013. Short-term observations of marine bacterial and viral communities: patterns,connections and resilience.e ISME Journal 7:1274–1285.Nemergut D.R., Schmidt S.K., Fukami T., O’Neill S.P., Bilinski T.M., Stanish L.F.,Knelman J.E., Darcy J.L., Lynch R.C., Wickey P., and Ferrenberg S. 2013. Patterns andprocesses of microbial community assembly. Microbiology and Molecular BiologyReviews 77:342–356.Ng T.F.F., Marine R., Wang C., Simmonds P., Kapusinszky B., Bodhidatta L., OderindeB.S., Wommack K.E., and Delwart E. 2012. High variety of known and new RNAand DNA viruses of diverse origins in untreated sewage. Journal of Virology 86:12161–12175.Nissimov J.I., Napier J.A., Allen M.J., and Kimmance S.A. 2016. Intragenus competitionbetween coccolithoviruses: an insight on how a select few can come to dominatemany.Environmental Microbiology 18:133–145.Noble R.T. and Fuhrman J.A. 1997. Virus decay and its causes in coastal waters. Appliedand Environmental Microbiology 63:77–83.Oda T., Nakamura A., Okamoto T., Ishimatsu A., and Muramatsu T. 1998. Lectin-induced enhancement of superoxide anion production by red tide phytoplankton.Marine Biology 131:383–390.Oksanen J., Blanchet F.G., Kindt R., Legendre P., Minchin P.R., O’Hara R.B., SimpsonG.L., Solymos P., Stevens M.H.H., and Wagner H. 2013. vegan: Community EcologyPackage. R package version 2.0-8.Oksanen J., Blanchet F.G., Kindt R., Legendre P., Minchin P.R., O’Hara R.B., SimpsonG.L., Solymos P., Stevens M.H.H., and Wagner H. 2015. vegan: Community EcologyPackage. R package version 2.3-0.Ono K., Khan S., and Onoue Y. 2000. Ešects of temperature and light intensity onthe growth and toxicity of Heterosigma akashiwo (Raphidophyceae). AquacultureResearch 31:427–433.Ono K., Muetze T., Kolishovski G., Shannon P., and Demchak B. 2015. CyREST:Turbocharging Cytoscape Access for External Tools via a RESTful API [version 1;referees: 2 approved]. F1000Research 4.190bibliographyPagarete A., ChowC.E.T., Johannessen T., Fuhrman J.A.,ingstad T.F., and Sandaa R.A.2013. Strong seasonality and interannual recurrence inmarinemyovirus communities.Applied and Environmental Microbiology 79:6253–6259.Parada A., Needham D.M., and Fuhrman J.A. 2016. Every base matters: assessing smallsubunit rRNA primers for marine microbiomes with mock communities, time-seriesand global eld samples. Environmental Microbiology 18:1403–1414.Paradis E. 2012. Analysis of Phylogenetics and Evolution with R. Use R!, Springer, NewYork.Parsons R.J., Breitbart M., Lomas M.W., and Carlson C.A. 2012. Ocean time-seriesreveals recurring seasonal patterns of virioplankton dynamics in the northwesternSargasso Sea.e ISME journal 6:273–284.Parsons T., Maita Y., and Lalli C. 1984. Amanual of chemical and biological methods forseawater analysis. Pergamon international library of science, technology, engineering,and social studies, Pergamon Press, New York.Paul J.H. 2008. Prophages in marine bacteria: dangerous molecular time bombs or thekey to survival in the seas?e ISME Journal 2:579–589.Payet J. and Suttle C. 2014. Viral infection of bacteria and phytoplankton in the ArcticOcean as viewed through the lens of ngerprint analysis. Aquatic Microbial Ecology72:47–61.Pedrós-Alió C. 2012. e rare bacterial biosphere. Annual Review of Marine Science4:449–466.Petraitis P.S., Latham R.E., and Niesenbaum R.A. 1989. e maintenance of speciesdiversity by disturbance.e Quarterly Review of Biology 64:393–418.Peura S., Bertilsson S., Jones R.I., and Eiler A. 2015. Resistant microbial cooccurrencepatterns inferred by network topology. Applied and Environmental Microbiology81:2090–2097.Philippe N., Legendre M., Doutre G., Couté Y., Poirot O., Lescot M., Arslan D., SeltzerV., Bertaux L., Bruley C., Garin J., Claverie J.M., and Abergel C. 2013. Pandoraviruses:amoeba viruses with genomes up to 2.5 Mb reaching that of parasitic eukaryotes.Science 341:281–286.PielouE.C. 1966.emeasurement of diversity in dišerent types of biological collections.Journal ofeoretical Biology 13:131 – 144.Pinhassi J., Sala M.M., Havskum H., Peters F., Guadayol O.S., Malits A., and MarraseC. 2004. Changes in bacterioplankton composition under dišerent phytoplanktonregimens. Applied and Environmental Microbiology 70:6753–6766.Pinto A.J. and Raskin L. 2012. PCR biases distort bacterial and archaeal communitystructure in pyrosequencing datasets. PLoS ONE 7:e43093.191bibliographyPowers L., Creed I.F., and Trick C.G. 2012. Sinking of Heterosigma akashiwo results inincreased toxicity of this harmful algal bloom species. Harmful Algae 13:95–104.Price M.N., Dehal P.S., and Arkin A.P. 2010. FastTree 2–approximately maximum-likelihood trees for large alignments. PloS one 5:e9490.Proctor L.M. and Fuhrman J.A. 1990. Viral mortality of marine bacteria and cyanobac-teria. Nature 343:60–62.Qiu X., Wu L., Huang H., McDonel P.E., Palumbo A.V., Tiedje J.M., and Zhou J. 2001.Evaluation of PCR-Generated chimeras, mutations, and heteroduplexes with 16SrRNA gene-based cloning. Applied and Environmental Microbiology 67:880–887.Quast C., Pruesse E., Yilmaz P., Gerken J., Schweer T., Yarza P., Peplies J., and GlocknerF.O. 2013.e SILVA ribosomal RNAgene database project: improved data processingand web-based tools. Nucleic Acids Research 41:D590–D596.Quince C., Lanzén A., Curtis T.P., Davenport R.J., Hall N., Head I.M., Read L.F.,and Sloan W.T. 2009. Accurate determination of microbial diversity from 454pyrosequencing data. Nature Methods 6:639–641.R Core Team. 2014. R: A language and environment for statistical computing. RFoundation for Statistical Computing, Vienna, Austria.Rambaut A. 2014. FigTree. http://tree.bio.ed.ac.uk/soŸware/gtree/ Version 1.4.2.Reeder J. and Knight R. 2010. Rapidly denoising pyrosequencing amplicon reads byexploiting rank-abundance distributions. Nature Methods 7:668–669.Renzette N., Pokalyuk C., Gibson L., Bhattacharjee B., Schleiss M.R., Hamprecht K.,Yamamoto A.Y., Mussi-Pinhata M.M., Britt W.J., Jensen J.D., and Kowalik T.F. 2015.Limits and patterns of cytomegalovirus genomic diversity in humans. Proceedings ofthe National Academy of Sciences 112:E4120–E4128.Reynolds C.S., Padisák J., and Sommer U. 1993. Intermediate disturbance in the ecologyof phytoplankton and themaintenance of species diversity: a synthesis. Hydrobiologia249:183–188.Rho M., Tang H., and Ye Y. 2010. FragGeneScan predicting genes in short and error-prone reads. Nucleic Acids Research 38:e191–e191.Rodriguez-Brito B., Li L., Wegley L., Furlan M., Angly F., Breitbart M., Buchanan J.,Desnues C., Dinsdale E., Edwards R., Felts B., Haynes M., Liu H., Lipson D., MahašyJ., Martin-CuadradoA.B.,MiraA., Nulton J., Pašić L., Rayhawk S., Rodriguez-MuellerJ., Rodriguez-Valera F., Salamon P., Srinagesh S.,ingstad T.F., Tran T.,urber R.V.,Willner D., Youle M., and Rohwer F. 2010. Viral and microbial community dynamicsin four aquatic environments.e ISME Journal 4:739–751.192bibliographyRodriguez-Valera F., Martin-CuadradoA.B., Rodriguez-Brito B., Pašić L.,ingstad T.F.,Rohwer F., and Mira A. 2009. Explaining microbial population genomics throughphage predation. Nature Reviews Microbiology 7:828–836.Romano C.M., Lauck M., Salvador F.S., Lima C.R., Villas-Boas L.S., Araújo E.S.A., LeviJ.E., Pannuti C.S., O’Connor D., and Kallas E.G. 2013. Inter- and intra-host viraldiversity in a large seasonal DENV2 outbreak. PLoS ONE 8:e70318.Rosenzweig M.L. 1995. Species diversity in space and time. Cambridge University Press.Rosetta C.H. and McManus G.B. 2003. Feeding by ciliates on two harmful algal bloomspecies, Prymnesium parvum and Prorocentrum minimum. Harmful Algae 2:109–126.Rozon R.M. and Short S.M. 2013. Complex seasonality observed amongst diversephytoplankton viruses in the Bay of Quinte, an embayment of Lake Ontario.Freshwater Biology 58:2648–2663.Ruan Q., Dutta D., Schwalbach M.S., Steele J.A., Fuhrman J.A., and Sun F. 2006.Local similarity analysis reveals unique associations among marine bacterioplanktonspecies and environmental factors. Bioinformatics 22:2532–2538.Sandaa R.A., Gómez-Consarnau L., Pinhassi J., Riemann L., Malits A., WeinbauerM.G., Gasol J.M., and ingstad T.F. 2009. Viral control of bacterial biodiversity -evidence from a nutrient-enriched marine mesocosm experiment. EnvironmentalMicrobiology 11:2585–2597.Sanfaçon H., Wellink J., Gall O., Karasev A., Vlugt R., and Wetzel T. 2009. Secoviridae:a proposed family of plant viruses within the order Picornavirales that combinesthe families Sequiviridae and Comoviridae, the unassigned genera Cheravirus andSadwavirus, and the proposed genus Torradovirus. Archives of Virology 154:899–907.Sanjuan R., Nebot M.R., Chirico N., Mansky L.M., and Belshaw R. 2010. Viral mutationrates. Journal of Virology 84:9733–9748.Schirmer M., Ijaz U.Z., D’Amore R., Hall N., Sloan W.T., and Quince C. 2015. Insightinto biases and sequencing errors for amplicon sequencing with the Illumina MiSeqplatform. Nucleic Acids Research 43:e37.Schloss P.D., GeversD., andWestcott S.L. 2011. Reducing the ešects of PCR amplicationand sequencing artifacts on 16s rRNA-based studies. PLoS ONE 6:e27310.Schloss P.D., Westcott S.L., Ryabin T., Hall J.R., HartmannM., Hollister E.B., LesniewskiR.A., Oakley B.B., Parks D.H., Robinson C.J., Sahl J.W., Stres B., allinger G.G.,Van Horn D.J., and Weber C.F. 2009. Introducing mothur: open-source, platform-independent, community-supported soŸware for describing and comparing micro-bial communities. Applied and Environmental Microbiology 75:7537–7541.Schluter D. 1984. A variance test for detecting species associations, with some exampleapplications. Ecology 65:998–1005.193bibliographySchmidt T.S.B., Matias Rodrigues J.F., and vonMering C. 2015. Limits to robustness andreproducibility in the demarcation of operational taxonomic units: Robustness andreproducibility in OTU demarcation. Environmental Microbiology 17:1689–1706.Schroeder D.C., Oke J., Hall M., Malin G., and Wilson W.H. 2003. Virus succession ob-served during an Emiliania huxleyi bloom. Applied and Environmental Microbiology69:2484–2490.Schwalbach M.S., Hewson I., and Fuhrman J.A. 2004. Viral ešects on bacterialcommunity composition inmarine planktonmicrocosms. AquaticMicrobial Ecology34:117–127.Seymour J.R., Seuront L., Doubell M., Waters R.L., and Mitchell J.G. 2006. Microscalepatchiness of virioplankton. Journal of the Marine Biological Association of theUnited Kingdom 86:551–562.Shade A., Jones S.E., Caporaso J.G., Handelsman J., Knight R., Fierer N., and GilbertJ.A. 2014. Conditionally rare taxa disproportionately contribute to temporal changesin microbial diversity. mBio 5:e01371–14–e01371–14.Shade A., Peter H., Allison S.D., Baho D.L., Berga M., Bürgmann H., Huber D.H.,Langenheder S., Lennon J.T., Martiny J.B.H., Matulich K.L., Schmidt T.M., andHandelsman J. 2012a. Fundamentals ofmicrobial community resistance and resilience.Frontiers in Microbiology 3.Shade A., Read J.S., Youngblut N.D., Fierer N., Knight R., Kratz T.K., Lottig N.R., RodenE.E., Stanley E.H., and Stombaugh J. 2012b. Lake microbial communities are resilientaŸer a whole-ecosystem disturbance.e ISME journal 6:2153–2167.Shannon C. 1948. A mathematical theory of distribution. e Bell System TechnicalJournal 27:623.Sher D., ompson J.W., Kashtan N., Croal L., and Chisholm S.W. 2011. Responseof Prochlorococcus ecotypes to co-culture with diverse marine bacteria. e ISMEjournal 5:1125–1132.Shirai Y., Tomaru Y., Takao Y., Suzuki H., Nagumo T., and Nagasaki K. 2008. Isolationand characterization of a single-stranded RNA virus infecting the marine planktonicdiatom Chaetoceros tenuissimusMeunier. Applied and Environmental Microbiology74:4022–4027.Short C.M., Rusanova O., and Short S.M. 2010. Quantication of virus genes providesevidence for seed-bank populations of phycodnaviruses in LakeOntario, Canada.eISME Journal 5:810–821.Short C.M. and Suttle C.A. 2005. Nearly identical bacteriophage structural genesequences are widely distributed in both marine and freshwater environments.Applied and Environmental Microbiology 71:480–486.194bibliographyShort S.M. 2012.e ecology of viruses that infect eukaryotic algae: Algal virus ecology.Environmental Microbiology 14:2253–2271.Short S.M. and Short C.M. 2009. Quantitative PCR reveals transient and persistent algalviruses in Lake Ontario, Canada. Environmental Microbiology 11:2639–2648.Short S.M. and Suttle C.A. 2002. Sequence analysis of marine virus communities revealsthat groups of related algal viruses are widely distributed in nature. Applied andEnvironmental Microbiology 68:1290–1296.Short S.M. and Suttle C.A. 2003. Temporal dynamics of natural communities of marinealgal viruses and eukaryotes. Aquatic Microbial Ecology 32:107–119.Sievers F., Wilm A., Dineen D., Gibson T.J., Karplus K., Li W., Lopez R., McWilliamH., Remmert M., Soding J., ompson J.D., and Higgins D.G. 2014. Fast, scalablegeneration of high-quality protein multiple sequence alignments using ClustalOmega. Molecular Systems Biology 7:539–539.Simon M., López-García P., Deschamps P., Moreira D., Restoux G., Bertolino P.,and Jardillier L. 2015. Marked seasonality and high spatial variability of protistcommunities in shallow freshwater systems.e ISME Journal 9:1941–1953.Simpson E. 1949. Measurement of diversity. Nature 163:688–688.Sipos R., Székely A.J., Palatinszky M., Révész S., Márialigeti K., and Nikolausz M. 2007.Ešect of primer mismatch, annealing temperature and PCR cycle number on 16srRNA gene-targetting bacterial community analysis. FEMS Microbiology Ecology60:341–350.Smoot M.E., Ono K., Ruscheinski J., Wang P.L., and Ideker T. 2011. Cytoscape 2.8: newfeatures for data integration and network visualization. Bioinformatics 27:431–432.Sogin M.L., Morrison H.G., Huber J.A., Welch D.M., Huse S.M., Neal P.R., Arrieta J.M.,andHerndl G.J. 2006.Microbial diversity in the deep sea and the underexplored “rarebiosphere”. Proceedings of the National Academy of Sciences 103:12115–12120.Sorensen G., Baker A.C., Hall M.J., Munn C.B., and Schroeder D.C. 2009. Novel virusdynamics in an Emiliania huxleyi bloom. Journal of Plankton Research 31:787–791.Spatharis S., Tsirtsis G., Danielidis D.B., Chi T.D., andMouillot D. 2007. Ešects of pulsednutrient inputs on phytoplankton assemblage structure and blooms in an enclosedcoastal area. Estuarine, Coastal and Shelf Science 73:807–815.Spencer R. 1955. A marine bacteriophage. Nature 175:690–691.Spencer R. 1960. Indigenous marine bacteriophages. Journal of Bacteriology 79:614.Srivastava D.S., Cadotte M.W., MacDonald A.A.M., Marushia R.G., and MirotchnickN. 2012. Phylogenetic diversity and the functioning of ecosystems. Ecology Letters15:637–648.195bibliographyStackebrandt E. and Goebel B.M. 1994. Taxonomic note: a place for DNA-DNAreassociation and 16s rRNA sequence analysis in the present species denition inbacteriology. International Journal of Systematic Bacteriology 44:846–849.Stamatakis A. 2014. RAxML version 8: a tool for phylogenetic analysis and post-analysisof large phylogenies. Bioinformatics 30:1312–1313.Stamatakis A., Hoover P., and Rougemont J. 2008. A rapid bootstrap algorithm for theRAxML web servers. Systematic Biology 57:758–771.Steele J.A., Countway P.D., Xia L., Vigil P.D., Beman J.M., Kim D.Y., Chow C.E.T.,Sachdeva R., Jones A.C., and Schwalbach M.S. 2011. Marine bacterial, archaeal andprotistan association networks reveal ecological linkages.e ISME journal 5:1414–1425.Steward G.F., Culley A.I., Mueller J.A., Wood-Charlson E.M., Belcaid M., and PoissonG. 2012. Are we missing half of the viruses in the ocean?e ISME Journal 7:672–679.Steward G.F., Montiel J.L., and Azam F. 2000. Genome size distributions indicate vari-ability and similarities among marine viral assemblages from diverse environments.Limnology and Oceanography 45:1697–1706.Storesund J.E., Erga S.R., Ray J.L.,ingstad T.F., and Sandaa R.A. 2015. Top-down andbottom-up control on bacterial diversity in a western Norwegian deep-silled 9ord.FEMS Microbiology Ecology 91:v076.Storey J.D., XiaoW., Leek J.T., Tompkins R.G., andDavis R.W. 2005. Signicance analysisof time course microarray experiments. Proceedings of the National Academy ofSciences of the United States of America 102:12837–12842.SullivanM.B. 2015. Viromes, not genemarkers, for studying double-strandedDNAviruscommunities. Journal of Virology 89:2459–2461.Sullivan M.B., Waterbury J.B., and Chisholm S.W. 2003. Cyanophages infecting theoceanic cyanobacterium Prochlorococcus. Nature 424:1047–1051.Suttle C.A. 2005. Viruses in the sea. Nature 437:356–361.Suttle C.A. 2007. Marine viruses - major players in the global ecosystem. Nat Rev Micro5:801–812.Suttle C.A. and Chan A.M. 1994. Dynamics and distribution of cyanophages andtheir ešect on marine Synechococcus spp. Applied and Environmental Microbiology60:3167–3174.Suttle C.A., Chan A.M., and Cottrell M.T. 1990. Infection of phytoplankton by virusesand reduction of primary productivity. Nature 347:467–469.Suttle C.A., Chan A.M., and Cottrell M.T. 1991. Use of ultraltration to isolateviruses from seawater which are pathogens of marine phytoplankton. Applied andEnvironmental Microbiology 57:721–726.196bibliographySuttle C.A. and Chen F. 1992. Mechanisms and rates of decay of marine viruses inseawater. Applied and Environmental Microbiology 58:3721–3729.Tai V., Lawrence J.E., Lang A.S., Chan A.M., Culley A.I., and Suttle C.A. 2003. Char-acterization of HaRNAV, a single-stranded RNA virus causing lysis of Heterosigmaakashiwo (Raphidophyceae). Journal of Phycology 39:343–352.Takao Y., Nagasaki K., Mise K., Okuno T., and Honda D. 2005. Isolation and charac-terization of a novel single-stranded rna virus infectious to a marine fungoid protist,Schizochytrium sp. (raustochytriaceae, Labyrinthulea). Applied and EnvironmentalMicrobiology 71:4516–4522.Tan J., Kelly C.K., and Jiang L. 2013. Temporal niche promotes biodiversity duringadaptive radiation. Nature Communications 4.Tarutani K., Nagasaki K., and Yamaguchi M. 2000. Viral impacts on total abundanceand clonal composition of the harmful bloom-forming phytoplankton Heterosigmaakashiwo. Applied and Environmental Microbiology 66:4916–4920.Taylor F. and Haigh R. 1993. e ecology of sh-killing blooms of the chloromonadžagellate Heterosigma in the Strait of Georgia and adjacent waters. In ToxicPhytoplankton Blooms in the Sea, pages 705–710.Taylor F., Haigh R., and Sutherland T. 1994. Phytoplankton ecology of Sechelt Inlet, a9ord system on the British Columbia coast. II. Potentially harmful species. MarineEcology Progress Series 104:151–164.te Velthuis A.J.W. 2014. Common and unique features of viral RNA-dependentpolymerases. Cellular and Molecular Life Sciences 71:4403–4420.Teeling H., Fuchs B.M., Becher D., KlockowC., Gardebrecht A., Bennke C.M., KassabgyM., Huang S., Mann A.J., Waldmann J., Weber M., Klindworth A., Otto A., Lange J.,Bernhardt J., Reinsch C., Hecker M., Peplies J., Bockelmann F.D., Callies U., GerdtsG., Wichels A., Wiltshire K.H., Glöckner F.O., Schweder T., and Amann R. 2012.Substrate-controlled succession of marine bacterioplankton populations induced bya phytoplankton bloom. Science 336:608–611.ingstad T.F. 2000. Elements of a theory for the mechanisms controlling abundance,diversity, and biogeochemical role of lytic bacterial viruses in aquatic systems.Limnology and Oceanography 45:1320–1328.ingstad T.F., Pree B., Giske J., and Våge S. 2015.What dišerence does it make if virusesare strain-, rather than species-specic? Frontiers in Microbiology 6:320.ingstad T.F., Våge S., Storesund J.E., Sandaa R.A., and Giske J. 2014. A theoreticalanalysis of how strain-specic viruses can control microbial species diversity.Proceedings of the National Academy of Sciences 111:7813–7818.197bibliographyTomaru Y. 2015. Marine diatom viruses and their hosts: Resistance mechanisms andpopulation dynamics. Perspectives in Phycology 2:69–81.Tomaru Y., Katanozaka N., Nishida K., Shirai Y., Tarutani K., Yamaguchi M., andNagasaki K. 2004. Isolation and characterization of two distinct types of HcRNAV,a single-stranded RNA virus infecting the bivalve-killing microalga Heterocapsacircularisquama. Aquatic Microbial Ecology 34:207–218.Tomaru Y., Kimura K., and Nagasaki K. 2015. Marine protist viruses. InMarine Protists(S. Ohtsuka, T. Suzaki, T. Horiguchi, N. Suzuki, and F. Not, eds.), pages 501–517,Springer Japan, Tokyo.Tomaru Y., Takao Y., Suzuki H., Nagumo T., and Nagasaki K. 2009. Isolation andcharacterization of a single-stranded RNA virus infecting the bloom-forming diatomChaetoceros socialis. Applied and Environmental Microbiology 75:2375–2381.Tomaru Y., Toyoda K., Kimura K., Hata N., Yoshida M., and Nagasaki K. 2012. Firstevidence for the existence of pennate diatom viruses.e ISME Journal 6:1445–1448.Tucker C.M. and Cadotte M.W. 2013. Unifying measures of biodiversity: understandingwhen richness and phylogenetic diversity should be congruent. Diversity andDistributions 19:845–854.Van Etten J.L., Meints R.H., Burbank D.E., Kuczmarski D., Cuppels D.A., and LaneL.C. 1981. Isolation and characterization of a virus from the intracellular green algasymbiotic with Hydra viridis. Virology 113:704–711.Van Valen L. 1973. A new evolutionary law. Evolutionary theory 1:1–30.Venter J.C., Remington K., Heidelberg J.F., Halpern A.L., Rusch D., Eisen J.A., WuD., Paulsen I., Nelson K.E., Nelson W., Fouts D.E., Levy S., Knap A.H., LomasM.W., Nealson K., White O., Peterson J., Hošman J., Parsons R., Baden-Tillson H.,Pfannkoch C., Rogers Y.H., and Smith H.O. 2004. Environmental genome shotgunsequencing of the Sargasso Sea. Science 304:66–74.Verity P.G. 1987. Abundance, community composition, size distribution, and productionrates of tintinnids in Narragansett Bay, Rhode Island. Estuarine, Coastal and ShelfScience 24:671–690.Verity P.G. and Stoecker D. 1982. Ešects of Olisthodiscus luteus on the growth andabundance of tintinnids. Marine Biology 72:79–87.Våge S., Storesund J.E., andingstad T.F. 2013. SAR11 viruses and defensive host strains.Nature 499:E3–E4.Wang K., Wommack K.E., and Chen F. 2011. Abundance and distribution of Syne-chococcus spp. and cyanophages in the Chesapeake Bay. Applied and EnvironmentalMicrobiology 77:7459–7468.198bibliographyWang Q., Garrity G.M., Tiedje J.M., and Cole J.R. 2007. Naïve Bayesian classier forrapid assignment of rRNA sequences into the new bacterial taxonomy. Applied andEnvironmental Microbiology 73:5261–5267.Waterbury J.B. and Valois F.W. 1993. Resistance to co-occurring phages enables marineSynechococcus communities to coexist with cyanophages abundant in seawater.Applied and Environmental Microbiology 59:3393–3399.Watson S.J., Welkers M.R.A., Depledge D.P., Coulter E., Breuer J.M., de Jong M.D., andKellam P. 2013. Viral population analysis and minority-variant detection using shortread next-generation sequencing. Philosophical Transactions of the Royal Society B:Biological Sciences 368:20120205.Weinbauer M.G. 2004. Ecology of prokaryotic viruses. FEMS Microbiology Reviews28:127–181.Weitz J.S., Hartman H., and Levin S.A. 2005. Coevolutionary arms races betweenbacteria and bacteriophage. Proceedings of the National Academy of Sciences102:9535–9540.Weitz J.S., Stock C.A., Wilhelm S.W., Bourouiba L., Coleman M.L., Buchan A., FollowsM.J., Fuhrman J.A., Jover L.F., Lennon J.T., Middelboe M., Sonderegger D.L., SuttleC.A., Taylor B.P., Fredeingstad T., Wilson W.H., and Eric Wommack K. 2015. Amultitrophic model to quantify the ešects of marine viruses on microbial food websand ecosystem processes. ISME J 9:1352–1364.Weitz J.S. and Wilhelm S.W. 2012. Ocean viruses and their ešects on microbialcommunities and biogeochemical cycles. F1000 Biol Rep 4:17.Wickham H. 2009. ggplot2. Springer New York, New York, NY.Wilhelm S.W. and Suttle C.A. 1999. Viruses and nutrient cycles in the sea. BioScience49:781–788.Wilhelm S.W., Weinbauer M.G., Suttle C.A., and Ješrey W.H. 1998.e role of sunlightin the removal and repair of viruses in the sea. Limnology and Oceanography 43:586–592.Wilkinson L. 2011. venneuler: Venn and Euler Diagrams. R package version 1.1-0.Williams R.J., Howe A., and Hofmockel K.S. 2014. Demonstrating microbial co-occurrence pattern analyses within and between ecosystems. Frontiers in Microbiol-ogy 5.Williamson S.J., Rusch D.B., Yooseph S., Halpern A.L., Heidelberg K.B., Glass J.I.,Andrews-Pfannkoch C., Fadrosh D., Miller C.S., Sutton G., Frazier M., and VenterJ.C. 2008. e Sorcerer II Global Ocean Sampling Expedition: metagenomiccharacterization of viruses within aquatic microbial samples. PLoS ONE 3:e1456.199bibliographyWilson W., Tarran G., Schroeder D., Cox M., Oke J., and Malin G. 2002a. Isolationof viruses responsible for the demise of an Emiliania huxleyi bloom in the EnglishChannel. Journal of the Marine Biological Association of the UK 82:369–377.Wilson W., Tarran G., and Zubkov M.V. 2002b. Virus dynamics in a coccolithophore-dominated bloom in the North Sea. Deep Sea Research Part II: Topical Studies inOceanography 49:2951–2963.WingetD.M., HeltonR.R.,WilliamsonK.E., Bench S.R.,Williamson S.J., andWommackK.E. 2011. Repeating patterns of virioplankton production within an estuarineecosystem. Proceedings of the National Academy of Sciences 108:11506–11511.Winget D.M. and Wommack K.E. 2008. Randomly amplied polymorphic DNA PCRas a tool for assessment of marine viral richness. Applied and EnvironmentalMicrobiology 74:2612–2618.Winter C., Bouvier T., Weinbauer M.G., andingstad T.F. 2010. Trade-ošs betweencompetition and defense specialists among unicellular planktonic organisms: the"Killing the Winner" hypothesis revisited. Microbiology and Molecular BiologyReviews 74:42–57.Woese C.R. and Fox G.E. 1977. Phylogenetic structure of the prokaryotic domain: theprimary kingdoms. Proceedings of the National Academy of Sciences 74:5088–5090.Woese C.R., Kandler O., andWheelis M.L. 1990. Towards a natural system of organisms:proposal for the domainsArchaea, Bacteria, and Eucarya. Proceedings of theNationalAcademy of Sciences 87:4576–4579.Worden A.Z., Follows M.J., Giovannoni S.J., Wilken S., Zimmerman A.E., and KeelingP.J. 2015. Rethinking the marine carbon cycle: Factoring in the multifarious lifestylesof microbes. Science 347:1257594–1257594.Xia L., Steele J., Cram J., Cardon Z., Simmons S., Vallino J., Fuhrman J., and Sun F. 2011.Extended local similarity analysis (eLSA) of microbial community and other timeseries data with replicates. BMC Systems Biology 5:S15.Yang Y., Motegi C., Yokokawa T., and Nagata T. 2010. Large-scale distribution patternsof virioplankton in the upper ocean. Aquatic Microbial Ecology 60:233–246.Yokoyama R. and Honda D. 2007. Taxonomic rearrangement of the genusSchizochytrium sensu lato based on morphology, chemotaxonomic characteristics,and 18S rRNA gene phylogeny (raustochytriaceae, Labyrinthulomycetes):emendation for Schizochytrium and erection of Aurantiochytrium andOblongichytrium gen. nov. Mycoscience 48:199–211.Yu G., Smith D., Zhu H., Guan Y., and Lam T.T.Y. 2016. ggtree: an R package forvisualization and annotation of phylogenetic tree with dišerent types of meta-data,submitted to Methods in Ecology and Evolution.200bibliographyZhang J. and Byrne C. 1999. Dišerential priming of RNA templates during cDNA syn-thesis markedly ašects both accuracy and reproducibility of quantitative competitivereverse-transcriptase PCR. Biochem. J 337:231–241.Zhang J., Kobert K., Flouri T., and StamatakisA. 2014. PEAR: a fast and accurate IlluminaPaired-End reAd mergeR. Bioinformatics 30:614–620.ZhaoY., TempertonB.,rash J.C., SchwalbachM.S., VerginK.L., LandryZ.C., EllismanM., Deerinck T., Sullivan M.B., and Giovannoni S.J. 2013. Abundant SAR11 viruses inthe ocean. Nature 494:357–360.Zhong X. and Jacquet S. 2014. Dišering assemblage composition and dynamics in T4-like myophages of two neighbouring sub-alpine lakes. Freshwater Biology 59:1577–1595.Zhou J., Wu L., Deng Y., Zhi X., Jiang Y.H., Tu Q., Xie J., Van Nostrand J.D., He Z.,and Yang Y. 2011. Reproducibility and quantitation of amplicon sequencing-baseddetection.e ISME journal 5:1303–1313.Zingone A., Dubroca L., Iudicone D., Margiotta F., Corato F., Ribera d’Alcalà M.,Saggiomo V., and Sarno D. 2009. Coastal phytoplankton do not rest in winter.Estuaries and Coasts 33:342–361.201appendix aSupplementary Information to Chapter 2a.1 supplementary data and figuresa.1 .1 Control libraries:Methods: — Two control libraries were prepared in addition to the 5 sample librariesprepared. To get the Sanger sequence for 1 amplicon, from one PCR amplication, asingle sequence was cloned into TOPO TA vector (Invitrogen) and transformed intoE. coli grown in LB + amp at 37○C overnight. Colony PCR was performed on severalclones. Positive PCRproducts were cleaned using theQiagenMinelute PCR cleanup andsequenced using theM13 forward primer at NAPS (UBC) on anABI sequencer using BigDye Chemistry. To make the unamplied control sequence the plasmid containing theclone was grown in large quantities overnight (6x75ml cultures). Cells were harvested bycentrifugation at 3200 g at 4○ for 20 min.e plasmids were extracted using the Qiagenmini-prep plasmid kit and digested with Ecor I for 2 h (600 µl DNA, 60 µl React 3 bušer,300 µl Ecor I, 38.5 µl H20) to cut out the product from the vector.e digested extractswere run on 1.5% agarose gels and the Qiagen Minelute gel extraction kit was to purifythe desired cut product. e puried cut product was processed in library preparationlike the other samples (see Chapter 2 Materials and Methods). To make the ampliedcontrol sequences, the puried digestedmaterial was used as template in a PCR reactionas detailed above and the product was used in library preparation. Alignments of thecontrol sequences were visualized in Geneious (v.6.1.6)(Kearse et al., 2012).202appendix aResults — e control libraries contained a total of 94 reads.ree reads were recov-ered from the non-amplied control sequence and 91 reads from the amplied clonedsequence. e non-amplied control sequences had no errors. However, there wereonly 3 reads recovered from that library. erefore, it is di›cult to compare to theamplied library. e 91 reads from the amplied library contained some sequenceswith insertions and some with erroneous base-calls (Figure a.2).ese control libraries enabled condent testing of the error-correction algorithm(Reeder and Knight, 2010). ere were errors such as homopolymers and insertionsattributable to PCR amplication and 454 pyrosequencing. However, the denoiser algo-rithm adequately corrected the viral OTU reads.203appendix aFigure a.1: Percent similarity vs. number of OTUs. All sequences were translated toamino acids using FragGeneScan with the 454_10 training option (Rho et al., 2010) andwere clustered with uclust (Edgar, 2010). 204appendix aFigure a.2: Control sequence reads clustered at 95% similarity. All sequences weretranslated to amino acids using FragGeneScan with the 454_10 training option (Rhoet al., 2010) and clustered with uclust at 95% similarity using centroids as the output(Edgar, 2010). Sequences were aligned using default parameters for MUSCLE(Edgar,2004). Mismatches in clustered sequences are highlighted in colours. Screenshot wastaken from Geneious (v.6.1.6)(Kearse et al., 2012).205appendix aFigurea.3: Control sequence andPCRamplied reads denoised at dišerent percentagesusing the QIIME denoiser Titanium settings (Reeder and Knight, 2010). All sequenceswere translated to amino acids using FragGeneScan with the 454_10 training option(Rho et al., 2010) and clusteredwith uclust at 95% similarity using centroids as the output(Edgar, 2010). Sequences were aligned using default parameters for MUSCLE (Edgar,2004). Screenshot was taken from Geneious (v.6.1.6) (Kearse et al., 2012).206appendix bSupplementary Information to Chapter 3b.1 supplementary figuresb.1 .1 Detailed phylogenetic treesFigure b.1: Legend for tip colours forMaximum likelihood trees formarine picorna-likeviruses.Marine Picorna-like viruses —T4-like myoviruses —b.1 .2 Heatmaps over timeEukaryotes — ere are no large scale žuctuations over time in the eukaryotic OTUs(Figure b.18). e OTUs that dominated the community žuctuated and were generallypresent, but some groups come and go a bit. Most notably the 1st and 2nd samples inJanuary seem to have the most shiŸs between them.Bacteria — In the bacterial OTUs many OTUs are persistent and present except in1-2 samples (Figure b.19). One sample in late June shows that there is a large turnoverin the whole community and the community is then dominated by one group of relatedbacteria.207appendix bFigure b.2: Maximum likelihood phylogenetic tree (RAxML) of subsection A of RdRpincluding reference sequences and OTUs generated in this study. Subsection views arefor the reference isolates (in black) and Group A (in grey). Outgroup is virus Equinerhinitis B virus (Picornaviridae). OTUs at 95% similarity at the amino-acid level.208appendix bFigure b.3: Maximum likelihood phylogenetic tree (RAxML) of subsections B and Cof RdRp including reference sequences and OTUs generated in this study. Subsectionviews are for the Group B (in blue) and Group C (in orange). Outgroup is virus Equinerhinitis B virus (Picornaviridae). OTUs at 95% similarity at the amino-acid level.209appendix bFigure b.4: Maximum likelihood phylogenetic tree (RAxML) of subsections D and Eof RdRp including reference sequences and OTUs generated in this study. Subsectionviews are for the GroupD (in purple) and Group E (in fushia). Outgroup is virus Equinerhinitis B virus (Picornaviridae). OTUs at 95% similarity at the amino-acid level.210appendix bFigure b.5: Maximum likelihood phylogenetic tree (RAxML) of subsection F and Gof RdRp including reference sequences and OTUs generated in this study. Subsectionviews are for the Group F (in green) and Group G (in yellow). Outgroup is virus Equinerhinitis B virus (Picornaviridae). OTUs at 95% similarity at the amino-acid level.211appendix bFigure b.6: Maximum likelihood phylogenetic tree (RAxML) of subsection H of RdRpincluding reference sequences and OTUs generated in this study. Subsection views arefor the Group H (in brown). Outgroup is virus Equine rhinitis B virus (Picornaviridae).OTUs at 95% similarity at the amino-acid level.212appendix bFigure b.7: Legend for tip colours for the Maximum likelihood tree of T4-likemyovirises.213appendix bFigure b.8: Maximum likelihood phylogenetic tree (RAxML) of subsection A of gp23(marker for T4-like myoviruses) including reference sequences and OTUs generated inthis study. Subsection views are for the reference isolates (in black) and Group A (ingrey). Outgroup is Enterobacteria phage T4. OTUs at 95% similarity at the amino acidlevel.214appendix bFigure b.9: Maximum likelihood phylogenetic tree (RAxML) of subsection B of gp23(marker for T4-like myoviruses) including reference sequences and OTUs generated inthis study. Subsection views are for Group B (in turquoise). Outgroup is Enterobacteriaphage T4. OTUs at 95% similarity at the amino acid level.215appendix bFigure b.10: Maximum likelihood phylogenetic tree (RAxML) of subsection C of gp23(marker for T4-like myoviruses) including reference sequences and OTUs generated inthis study. Subsection views are for Group C (in yellow). Outgroup is Enterobacteriaphage T4. OTUs at 95% similarity at the amino acid level.216appendix bFigure b.11: Maximum likelihood phylogenetic tree (RAxML) of subsection D andE of gp23 (marker for T4-like myoviruses) including reference sequences and OTUsgenerated in this study. Subsection views are for Group D (in purple) and Group E (inred). Outgroup is Enterobacteria phage T4. OTUs at 95% similarity at the amino acidlevel.217appendix bFigure b.12: Maximum likelihood phylogenetic tree (RAxML) of subsection F of gp23(marker for T4-like myoviruses) including reference sequences and OTUs generated inthis study. Subsection views are for Group F (in blue). Outgroup is Enterobacteria phageT4. OTUs at 95% similarity at the amino acid level.218appendix bFigure b.13: Maximum likelihood phylogenetic tree (RAxML) of subsection G (top)of gp23 (marker for T4-like myoviruses) including reference sequences and OTUsgenerated in this study. Subsection views are for the top portion of Group G (in orange).Outgroup is Enterobacteria phage T4. OTUs at 95% similarity at the amino acid level.219appendix bFigure b.14: Maximum likelihood phylogenetic tree (RAxML) of subsectionG (bottom)of gp23 (marker for T4-like myoviruses) including reference sequences and OTUsgenerated in this study. Subsection views are for the bottom portion of Group G (seeprevious sub tree for the top portion). Outgroup is Enterobacteria phage T4. OTUs at95% similarity at the amino acid level.220appendix bFigure b.15: Maximum likelihood phylogenetic tree (RAxML) of subsection H of gp23(marker for T4-like myoviruses) including reference sequences and OTUs generated inthis study. Subsection views are for Group H (in green). Outgroup is Enterobacteriaphage T4. OTUs at 95% similarity at the amino acid level.221appendix bFigure b.16: Maximum likelihood phylogenetic tree (RAxML) of subsection I (top)of gp23 (marker for T4-like myoviruses) including reference sequences and OTUsgenerated in this study. Subsection views are for the top portion of Group I (in pink).Outgroup is Enterobacteria phage T4. OTUs at 95% similarity at the amino acid level.222appendix bFigure b.17: Maximum likelihood phylogenetic tree (RAxML) of subsection I (bottom)of gp23 (marker for T4-like myoviruses) including reference sequences and OTUsgenerated in this study. Subsection views are for the bottom portion of Group I (seeprevious sub tree for the top portion). Outgroup is Enterobacteria phage T4. OTUs at95% similarity at the amino acid level.223appendix bFigure b.18: Heatmap of relative abundance of eukaryotic OTUs (97% similarity) overtime. Each column is a time point224appendix bFigure b.19: Heatmap of relative abundance of bacterial OTUs (97% similarity) overtime. Each column is a time point225appendix bFigure b.20: Heatmap of relative abundance of T4-like myoviral OTUs (95% similarityamino acid) over time ordered by phylogenetic tree tree (Figure b.3.7). Each column isa time point226appendix bFigure b.21: Heatmap of relative abundance of marine picorna-like OTUs (95%similarity amino acid) over time ordered by phylogenetic tree tree (Figure b.3.6). Eachcolumn is a time point227appendix bFigure b.22: Non-metric dimensional scaling (NMDS) plot of the microbial communi-ties coloured by season. A)NMDS of eukaryotic communities usingHellinger’s distance.Seasons are dened according to text. B) Bacterial(16S) communitiy NMDS. C) Marinepicorna-like viral communitiy NMDS.D) T4-likemyoviral communityNMDS. Coloursby season. Lines connect sequential sampling times.228appendix cSupplementary Information to Chapter 4c.1 supplementary figuresc.1 .1 Network simulations over timeSimulations were performed with the same number of nodes and edges as from theobserved networks.229appendix cFigure c.1: Overall network (includes OTUs from eukaryotic, bacterial, T4-likemyoviruses and environmental parameters) compared to simulated networks createdfrom samenumber of nodes and edges. RandomNetwork (RN)was generated accordingto the Erdos-Renyi model and the Scale-Free (SF) were generated according to theBarabasi-Albert model.e green vertical line corresponds to the annual spring bloomand grey lines correspond to divisions between seasons.230appendix cFigure c.2: Network of eukaryotic OTUs compared to simulated networks created fromsame number of nodes and edges. Random Network (RN) was generated according tothe Erdos-Renyimodel and the Scale-Free (SF)were generated according to the Barabasi-Albert model.e green vertical line corresponds to the annual spring bloom and greylines correspond to divisions between seasons.231appendix cFigure c.3: Network of bacterial OTUs compared to simulated networks created fromsame number of nodes and edges. Random Network (RN) was generated according tothe Erdos-Renyimodel and the Scale-Free (SF)were generated according to the Barabasi-Albert model.e green vertical line corresponds to the annual spring bloom and greylines correspond to divisions between seasons.232appendix cFigure c.4: Network of marine picorna-like viral OTUs compared to simulatednetworks created from same number of nodes and edges. Random Network (RN) wasgenerated according to the Erdos-Renyi model and the Scale-Free (SF) were generatedaccording to the Barabasi-Albert model. e green vertical line corresponds to theannual spring bloom and grey lines correspond to divisions between seasons.233appendix cFigure c.5: Network of bacterial and T4-like myoviral OTUs compared to simulatednetworks created from same number of nodes and edges. Random Network (RN) wasgenerated according to the Erdos-Renyi model and the Scale-Free (SF) were generatedaccording to the Barabasi-Albert model. e green vertical line corresponds to theannual spring bloom and grey lines correspond to divisions between seasons.234appendix cFigure c.6: Network of T4-like myoviral OTUs compared to simulated networkscreated from same number of nodes and edges. Random Network (RN) was generatedaccording to the Erdos-Renyi model and the Scale-Free (SF) were generated accordingto the Barabasi-Albert model. e green vertical line corresponds to the annual springbloom and grey lines correspond to divisions between seasons.235appendix cFigure c.7: Network of eukaryotic and bacterial OTUs compared to simulated networkscreated from same number of nodes and edges. Random Network (RN) was generatedaccording to the Erdos-Renyi model and the Scale-Free (SF) were generated accordingto the Barabasi-Albert model. e green vertical line corresponds to the annual springbloom and grey lines correspond to divisions between seasons.236appendix cFigure c.8: Network of eukaryotic and marine picorna-like viral OTUs compared tosimulated networks created from same number of nodes and edges. Random Network(RN) was generated according to the Erdos-Renyi model and the Scale-Free (SF) weregenerated according to the Barabasi-Albert model. e green vertical line correspondsto the annual spring bloom and grey lines correspond to divisions between seasons.237

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0300438/manifest

Comment

Related Items