INSIGHT REVIEW NATURE|Vol 437|15 September 2005|doi:10.1038/nature04160 22 Viruses in the sea Curtis A. Suttle1 Viruses exist wherever life is found. They are a major cause of mortality, a driver of global geochemical cycles and a reservoir of the greatest genetic diversity on Earth. In the oceans, viruses probably infect all living things, from bacteria to whales. They affect the form of available nutrients and the termination of algal blooms. Viruses can move between marine and terrestrial reservoirs, raising the spectre of emerging pathogens. Our understanding of the effect of viruses on global systems and processes continues to unfold, overthrowing the idea that viruses and virus-mediated processes are sidebars to global processes. For years viruses were known to exist in seawater, but reports 15 years ago caused great excitement by demonstrating not only that viruses are abundant, but that they infect the dominant organisms in the ocean1���3. These observations occurred against the backdrop of a major shift in thinking among oceanographers to acknowledge that bacteria and microbial processes are important players in the oceans. For example, because viruses are significant agents of microbial mortality, they have an effect on nutrient cycling4���6. Moreover, the narrow host range of most viruses suggests that infection is important in controlling the composition of planktonic communities6���8. Interest in the vast viral communities in the sea continues to expand as the relevance of viruses to evolution, pathogen emergence and even exobiology begins to be explored. Several excellent reviews4���7 have captured the advances in our understanding of marine viruses and their role in the ocean. This review focuses on areas where our knowledge is changing rapidly, where methodological problems have impeded progress and where new data are altering perceptions. Without doubt, viruses are the most abundant and genetically diverse ���life forms��� in the ocean. They are major pathogens of planktonic organisms and consequently are signif- icant players in nutrient and energy cycling. As well, they are pathogens of higher organisms and there is good evidence that some viruses move between marine and terrestrial reservoirs. Recognition that viruses play a major role in marine ecosystems has added a significant new dimen- sion to our understanding of biological oceanographic processes. Total virus abundance has been underestimated Viruses are extremely abundant in aquatic systems. The first observa- tions by transmission electron microscopy (TEM)1,2 indicated that, typ- ically, there are ~107 viruses ml���1 and that abundance decreases with depth and distance from the shore9,10. In general, abundance correlates with system productivity and is highest where bacteria and chlorophyll are greatest11,12. In marine sediments, abundances are even higher, with 108���109 viruses cm���3 typical in nearshore surface sediments1,10,13. Even 100 m below the sediment surface15 viruses can be plentiful, although at the sediment surface in the deep ocean they seem to be less abundant16. Epifluorescence microscopy (EM) is now the preferred method for counting viruses because of its higher accuracy and precision17,18, although flow cytometry shows promise as a high-throughput method19. The shift to EM-based techniques has not been without problems, including significant differences in results among methodologies20, con- cerns about reproducibility and effects of sample storage19,21. For instance, estimates of viral abundance performed using samples not immediately processed or frozen in liquid nitrogen can be an order of magnitude too low. Currently, our best estimates range from ~3��106 viruses ml���1 in the deep sea22,23 to ~108 viruses ml���1 in productive coastal waters. Assuming the volume of the oceans is 1.3��1021 l and the aver- age abundance of viruses is 3��109 l���1, then ocean waters contain ~4��1030 viruses. Because a marine virus contains about 0.2 fg of carbon and is about 100 nm long, this translates into 200 Mt of carbon in marine viruses. If the viruses were stretched end to end they would span ~10 million light years. In context, this is equivalent to the carbon in ~75 mil- lion blue whales (~10% carbon, by weight24), and is ~100 times the dis- tance across our own galaxy. This makes viruses the most abundant biological entity in the water column of the world���s oceans, and the sec- ond largest component of biomass after prokaryotes. But total viral abundance alone does not give us an indication of infectivity. Most viruses in seawater seem to be infectious25, and some can remain infectious in sediments for long periods, from decades to a hun- dred years or more26,27. Estimating the abundance of infectious viruses is complicated by strain specificity yet in offshore waters most collisions between bacteria and viruses seem to result in infection, suggesting that selection for resistance is low28. Even so, viruses infecting specific hosts can be extremely abundant. For example, viruses infecting single strains of the cyanobacterium Synechococcus28 or the photosynthetic flagellate Micromonas pusilla29 can occur in excess of 105 infectious units ml���1. Yet, even the most permissive hosts are not sensitive to infection by all viruses that infect a given species therefore, even these are underestimates. Ulti- mately, high strain specificity combined with poor representation in cul- ture of the dominant microbes in the sea means that the absolute abundance of infectious viruses must be deduced by inference. Unexplored diversity Virus form provides insight into function Not only are viruses abundant in oceans but, as is becoming clear, they also harbour enormous genetic and biological diversity. TEM studies 1Department of Chemistry, University of California, Berkeley and the Molecular Foundry, Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA.

NATURE|Vol 437|15 September 2005 INSIGHT REVIEW 23 Morphotype provides some insights into the selective pressures facing virus communities. Myoviruses with their broader host ranges can quickly take advantage of increases in host populations, consistent with r-selection (short generation times and high reproductive rates). By contrast, many siphoviruses can archive their genomes in host cells, tying their replication rate to that of the host, until an environmental cue triggers the lytic cycle. This suggests that siphoviruses are more K-selected (longer generation times and lower reproductive rates). There is much to learn about the biological and genetic diversity of marine phages. In culture, the phages used are unlikely to be repre- sentative of the dominant phages in the ocean, because the most abun- dant prokaryotes have proven very difficult to grow in culture, although this is beginning to change. The story with respect to viruses infecting eukaryotes is even more complex. The first virus isolates infecting eukaryotic phytoplankton belonged to a group of large double-stranded (ds) DNA viruses, the Phycodnaviridae, and included viruses that infected important taxa of marine primary producers including toxic bloom formers and macroalgae26,35. The scene is changing rapidly as the isolation of many previously unknown viruses (Table 1) greatly enriches the taxonomic and phylogenetic space of known viral ���life���. For instance, a single- stranded (ss)RNA virus that infects a toxic bloom-forming alga (Het- erosigma akashiwo) led to the creation of the Marnaviridae36, probably the first of many new marine virus families. Other examples include a previously unknown dsRNA virus that infects the photosynthetic fla- gellate, Micromonas pusilla37, a ssRNA virus that infects a thraus- tochytrid fungus38, and a nuclear-inclusion virus (NIV) that has both ss and dsDNA and infects the diatom alga Chaetoceros salsugineum39, and has virtually no similarity to known viruses. Other NIVs that infect H. akashiwo40 (Fig. 2) and Cheatoceros41, as well as a large dsDNA virus that infects a marine heterotrophic protist42, remain to be characterized. The largest virus genome belongs to Mimivirus43, which also infects a protist, whereas viruses that cause disease in shrimp44 and crabs45 are distantly related to other viruses, suggesting that heterotrophs will also yield an untapped oasis of unexplored viral diversity. Viral genetic diversity is extremely high Culture-independent approaches indicate that we are just scraping the surface of viral life in the oceans. No gene is found universally in viruses but some genes are representative of specific subsets of the viral community. The first studies used the DNA polymerase genes of Phy- codnaviridae46 to reveal enormous genetic variation that was not rep- resented in cultures, and showed that very similar sequences were present in distant oceans47. Even more striking results were obtained in studies targeting a subset of myoviruses: tremendous diversity occurred on large48 and small49 spatial scales, with most of the sequences falling into groups with no cultured representatives50,51. of marine viral communities1 and phage isolates30 reveal a plethora of morphotypes, whereas host-range studies show complex patterns of resistance and susceptibility31. Among marine phages (Fig. 1), those with contractile tails (such as myoviruses and T4-like viruses) and long flexible tails (such as siphoviruses and lambda-like viruses) are most frequently isolated32���34, even though TEM suggests that phages with short non-contractile tails (such as podoviruses and T7-like viruses) and without tails are most abundant. Because ���lifestyles��� among tailed phages differ, morphology provides clues about host range and viral replication. For example, myoviruses are typically lytic and often have a broader host range than other tailed phages, even infecting different species of bacteria33,34. By contrast, the host range of podoviruses is generally narrowest, with siphoviruses being intermediate. However, many siphoviruses can integrate into the host genome and be passed from generation to generation. Figure 1 | The three families of tailed dsDNA viruses (phages) that infect bacteria. a, Myoviruses are often the most commonly isolated phage from natural marine viral communities. They have contractile tails, are typically lytic and often have relatively broad host ranges. b, Podoviruses have a short non-contractile tail, are also typically lytic and have very narrow host ranges. They are less commonly isolated from seawater. c, Siphoviruses have long non-contractile tails. They are frequently isolated from seawater, often have a relatively broad host range, and many are capable of integrating into the host genome. Scale bar, 50 nm. Table 1 | Some unusual aquatic viruses whose genomic sequences have recently been completed Virus Family Nucleic acid Genome size (bp) Number of proteins Chaetoceros salsugineum Unassigned ssDNA / dsDNA 6,005 ssDNA nuclear inclusion virus39 997 dsDNA Unknown Emiliania huxleyi Phycodnaviridae dsDNA 407,339 472 virus86(ref. 63) Heterosigma akashiwo RNA Marnaviridae ssRNA 8,600 6 or 7 virus36 Micromonas pusilla37 Reoviridae dsRNA 26,000 Unknown on 11 segments Ectocarpus siliculosus Phycodnaviridae dsDNA 335,593 240 virus85 Mimivirus43 Mimiviridae dsDNA 1,181,404 911 White spot syndrome virus44 Nimaviridae dsDNA 305,107 531 All infect eukaryotic phytoplankton with the exception of the Ectocarpus, Mimivirus, and white-spot viruses, which infect a brown alga, freshwater protist and penaeid shrimp, respectively.