REVIEWS Marine microbial community dynamics and their ecological interpretation Jed A. Fuhrman, Jacob A. Cram and David M. Needham Abstract | Recent advances in studying the dynamics of marine microbial communities have shown that the composition of these communities follows predictable patterns and involves complex network interactions, which shed light on the underlying processes regulating these globally important organisms. Such ‘holistic’ (or organism- and system-based) studies of these communities complement popular reductionist, often culture-based, approaches for understanding organism function one gene or protein at a time. In this Review, we summarize our current understanding of marine microbial community dynamics at various scales, from hours to decades. We also explain how the data illustrate community resilience and seasonality, and reveal interactions among microorganisms. Phototrophic Marine microbial communities (consisting of bacteria, potentially to show which organisms interact and how. Organisms that can transform archaea, protists, fungi and viruses) process about one- In the context of this Review, the term dynamics is used light energy into biologically half of the global biogeochemical flux of biologically to refer to changes in the abundance of various mem- usable (chemical) forms. A important elements, such as carbon, nitrogen, phos- bers (or populations) in a community, but not necessar- photoautotroph obtains its 1–4 biomass carbon from fully phorus, sulphur and iron . These organisms include ily changes in the physiological state of these members; oxidized carbon (that is, phototrophic and chemotrophic primary producers, as the term community is used here to refer to the types of carbon dioxide) and a well as heterotrophic ‘secondary’ producers, which recy- microorganisms present and their relative proportions. photoheterotroph obtains its cle dissolved organic carbon and nutrients through the The time-series approach, which involves evaluating the biomass carbon from reduced microbial loop5. In recent years, molecular analysis has dynamics of microbial community composition and carbon (for example, organic matter). made great strides in determining the organisms that the surrounding environment over multiple time points are present at a given site and how they are distributed (BOX 1), including the subsequent analysis of the large Chemotrophic over space and time. Most of this information is derived data sets that are created (BOX 2; FIG. 1), can be considered Organisms that can obtain from phylogenetic analyses using a few informative a holistic, system-wide investigation that complements energy from chemical 6–8 reactions. A chemoautotroph genes, such as the 16S and 18S rRNA genes , but phy- the more reductionist approaches of examining the obtains its biomass carbon logenetic identification alone is insufficient to assess the system gene by gene or protein by protein. from carbon dioxide, and a environmental functions and ecology of the community. Recent studies of marine community dynamics over chemoheterotroph obtains its Although we are learning much about such functions via various timescales, from diel to interannual, have illu- biomass from reduced carbon. ‘omics’ (metagenomic, metatranscriptomic, metaprot- minated not only the environmental conditions that eomic and metabolomic analyses), such studies alone are preferred by individual microorganisms but also rarely provide the information that is needed to pre- the likely interactions among these organisms and the dict interactions, competition for nutrients, symbioses emergent properties of the system as a whole. In par- Department of Biological and other processes that determine the overall roles of ticular, these studies have shown that marine microbial Sciences and Wrigley Institute for Environmental Studies, the distinct organisms in the sea. Therefore, additional communities are dynamic (in a constant state of flux) University of Southern information, beyond that which can be derived from but also resilient, which means that their behaviours California, Los Angeles, omics, is needed to predict community functions and are broadly predictable in terms of typical features of California 90089–0371, USA. interactions between organisms. Towards this goal, daily, seasonal and interannual variation in community Correspondence to J.A.F. a great deal can be learned by evaluating the dynam- composition. This implies that despite external forces e-mail: [email protected] doi:10.1038/nrmicro3417 ics of community composition and the corresponding that alter the community (such as temperature, nutri- Published online environmental parameters to determine the extent to ent supply and physical mixing), there are internal feed- 9 February 2015 which such dynamics follow predictable patterns, and back mechanisms, including competition, viral infection NATURE REVIEWS | MICROBIOLOGY VOLUME 13 | MARCH 2015 | 133 © 2015 Macmillan Publishers Limited. All rights reserved REVIEWS Box 1 | Temporal dynamics versus spatial variation analyses are well developed and data sets are available. Owing to space limitations, an extensive discussion of All biological systems are dynamic on one or more scales. The term ‘dynamics’ protistan and phage dynamics is excluded, although specifically refers to changes over time, and temporal changes in microbial some prominent examples are highlighted. communities result from growth and death, as well as the import and export of each organism present in the community. Intraspecies evolution, resulting in the emergence What processes drive dynamics at different scales? of variants with new combinations of traits, may also contribute to community dynamics, but this is beyond the scope of this discussion. Microorganisms change over multiple timescales and in Dynamics are relatively straightforward in closed, artificial systems (such as response to different forces, including both biological bioreactors, in which import and export are constrained), but such closed systems are and non-biological properties of the environment that rare to non-existent in nature, and the sea is a particularly open and complex drive changes in microbial community composition. As environment. In marine systems, dynamics are complicated by water mixing and the typical average generation times of marine plankton advection via currents (which are directional) and eddies (which typically swirl). are approximately a day in surface waters, and longer in Sampling of dynamics in the sea is traditionally carried out by two alternative deep waters1, substantial changes in community compo- time-series sampling modes: Lagrangian, in which the sampling attempts to follow the sition in less than a few hours are not expected; there- prevailing current; and Eulerian, in which the sampler remains at a fixed geographic fore, this discussion addresses changes over timescales of location and currents will move water past the sampling location. Lagrangian sampling attempts to track a ‘parcel’ of water as it moves, but in practice even Lagrangian hours and longer. Important timescales and likely forc- sampling does not track a parcel of water because eddies mix the water and there are ing functions are outlined below, and these often overlap no truly stable parcels. Thus, the movement and mixing of water need to be considered with each other. when interpreting the dynamics of marine microbial communities to avoid confusing temporal dynamics with spatial variation. Most ocean time-series studies tend to use Hours. The physiological response of microorganisms Eulerian sampling for practical and logistical reasons; Lagrangian sampling is to changing conditions can occur rapidly, but composi- impractical in studies that are carried out over the course of more than a few weeks. tional changes observed at subdaily timescales can only It is also important to consider horizontal and vertical scales separately, because occur in communities in which organisms replicate rela- much of the ocean is stratified vertically, with restricted vertical mixing; thus, tively rapidly. Candidates for change on this timescale physicochemical and biological gradients can be on scales of metres (or less) vertically. are copiotrophic organisms (often Gammaproteobacteria, The extent to which water mixing (which drives the import and export of organisms) alters the composition of a parcel of water is determined by the composition of the Flavobacteriia, certain Alphaproteobacteria and others) adjacent water. Thus, although horizontal mixing occurs all the time (via diffusion and that can grow rapidly, and although these taxa are usu- eddies), a parcel of water can have a stable microbial composition as long as the ally rare in seawater, they can quickly become abun- microbiologically important conditions (for example, temperature and nutrient dant under suitable conditions9. Community changes concentrations) are consistent in the surrounding water. From the limited available that occur on the scale of hours might also result from data, it seems that the size of a typical microbiologically coherent parcel (with a rapid selective cell death, but this has not yet been doc- consistent microbial community composition) is in the order of 2–20 km in horizontal umented. Changes in community composition at this extent105,106. However, at fronts where conditions such as temperature or chlorophyll scale can result from both predictable and unpredict- levels change abruptly, sharper gradients in microbial composition