Patterns and Prediction in Microbial Oceanography

Microbes are integral to the structure and functioning of marine ecosystems, and to the chemistry of the ocean and its interaction with the atmosphere. They mediate the chemical transformations that over geological time have determined the composition of our atmosphere (Fennel et al., 2005) and the balance of major nutrients in the sea (Arrigo, 2005). These nutrients ultimately determine how much and what kinds of marine life can be supported in surface waters, the amount of primary production that can be transferred to higher trophic levels, and how much carbon is stored in the deep ocean (Falkowski et al., 1998). Microbes in the ocean are thus directly and indirectly sensitive to, and part of, the ocean’s response to global change. The manifestations of global change include not only the changes in ocean temperature, circulation, pH, and nutrient availability linked to increasing greenhouse gases, but also the profound alterations of the marine environment associated with a growing and rapidly developing human population—habitat destruction, coastal eutrophication, pollution (GESAMP, 2001), and the elimination of larger predators by fishing, leading to fundamental alterations in food webs (Pauly et al., 1998; Myers et al., 2007). Our challenge as scientists is to describe and understand the environmental control of marine microbial communities so that we can predict the influences of global change on the biology and chemistry of the ocean as it interacts with and influences the atmosphere and climate. The central questions are clear, and they share a common theme: How will marine microbes, and the ecosystems they support, respond to environmental change, and what are the consequences for ecosystems and the biosphere? Broadly, we need to describe and understand the structure (species composition, size distributions, and flows of energy and materials) and functioning (nature and efficiency of chemical transformations) of microbial communities, recognizing that structure and functioning are intimately linked.

its interaction with the atmosphere.
They mediate the chemical transformations that over geological time have determined the composition of our atmosphere (Fennel et al., 2005) and the balance of major nutrients in the sea (Arrigo, 2005). These nutrients ultimately determine how much and what kinds of marine life can be supported in surface waters, the amount of primary production that can be transferred to higher trophic levels, and how much carbon is stored in the deep ocean (Falkowski et al., 1998). Microbes in the ocean are thus directly and indirectly sensitive to, and part of, the ocean's response to global change. The manifestations of global change include not only the changes in ocean temperature, circulation, pH, and nutrient availability linked to increasing greenhouse gases, but also the profound alterations of the marine environment associated with a growing and rapidly developing human population-habitat destruction, coastal eutrophication, pollution (GESAMP, 2001), and the elimination of larger predators by fishing, leading to fundamental alterations in food webs (Pauly et al., 1998;Myers et al., 2007). Well-studied relationships include the Redfield ratios (Redfield, 1958), the Sheldon size spectrum (Sheldon et al., 1972;Platt and Denman, 1978), and patterns of phytoplankton abundance in relation to oceanographic influences (Sverdrup, 1955;Margalef et al., 1979;Longhurst, 2007) (Hutchinson, 1961), and invoked environmental variation as part of the explanation of coexistence. The work already described assumes a (marine) world in balance, but Hutchinson reminds us that local disturbances caused by everything from turbulence to marauding whales (highlighted by Smetacek and Nicol, 2005) complicate the simple picture.
This knowledge should change our view of marine communities fundamentally: if disturbance is superimposed upon the simplistic dynamic of species interactions, removing the competitive species that would otherwise dominate at a low-diversity equilibrium, then there will be selection for rapid growth under some conditions. Indeed, if within the Klausmeier et al. (2004aKlausmeier et al. ( , 2004b framework one optimizes allocation for rapid growth, selection favors lower N:P ratios, about half of the Redfield prediction. The equilibrium view predicts a ratio too high, and the growth-rate approach (see also Williams, 2006) (Droop, 1974) and Liebig's Law of the Minimum (which states simply that growth will be limited by whichever essential nutrient is in shortest supply), to show that, in general, the system's dynamics will reach an equilibrium in which a single factor is limiting. When this dynamic is embedded into an evolutionary framework, however, in which organisms must expend their available carbon on making nitrogen-rich proteins (the primary machines responsible for uptake of nutrients) or phosphorusrich ribosomes (which play a key role in growth), the system evolves to colimitation, in which neither nutrient is in excess. The reason for this is intuitively simple-if, for example, nitrogen alone were limiting, a mutant that relied less on nitrogen and made better use of the available phosphorus could proliferate.
Most importantly, the optimal allocation to proteins and ribosomes is determined primarily by the stoichiometry of proteins and ribosomes, and is largely independent of environmental conditions. Thus, at first blush, there is a justification for Redfield's interpretation.
On closer examination, however, some gaps remain. This simplistic approach predicts an N:P ratio that overestimates Redfield by a factor of three, primarily ...until recently, there was no basis for describing in a mechanistic way the ecological and biogeochemical links between processes on the molecular scale all the way to global change.
predicts one too low. This suggests that, in a spatio-temporal dynamic in which localized disturbances initiate temporal successions of species of increasing competitive ability but decreasing growth rates, coexistence will be possible and the model will predict the full range of empirically observed N:P ratios. On the regional scale, such as in the North Pacific subtropical gyre, the chemical composition of phytoplankton (and ultimately the N:P of the pools of dissolved and particulate organic matter in the surface layer) responds to relative supplies of N and P from below-that is, whichever runs out first. N 2 -fixing cyanobacteria, with inherently high N:P, are favored in oligotrophic waters with low N:P supply; their growth and subsequent influences on the chemical composition of deep water (increasing its N:P) lead to a hypothesized N 2 -fixation cycle between N-controlled and P-controlled systems (Karl, 2002)   Phytoplankton are patchy in their distributions, but zooplankton are even patchier because they actively aggregate.
Similarly, graphs of the relative abundances of all species assume canonical shapes for marine systems, just as they do for terrestrial systems, statistical regularities that summarize patterns of biodiversity. Most intriguing, perhaps, for marine systems, is the consistency in the power-law distributions that describe the size spectra, from the smallest microorganisms to the largest fish in the oceans ( Figure 1B, C). Sheldon et al. (1972) showed that the biomass spectrum in aquatic systems exhibits remarkable constancy in biomass in successive logarithmic size bins. Numerous explanations have been advanced for this regularity; although a universally satisfactory general explanation seems still lacking, these theoretical investigations have led to better understanding of the forces structuring marine ecosystems.
When sufficiently well developed, the hypotheses that explain macroscopic regularities can be used to describe and predict how ocean systems will respond to environmental change. An example is the Redfield-based description of the global balance between N 2 fixation and the microbially mediated loss of fixed nitrogen, as illustrated in Figure 2 and discussed by Arrigo (2005). This theoretical framework links environmentally determined controls on the growth rates of phytoplankton (i.e., replenishment of surface waters through upwelling or vertical mixing), phytoplankton N:P stoichiometry, and ultimately the N:P of the deep ocean. It can be exploited to hypothesize effects of another alteration Figure 1. Fundamental regularities in ocean biology and chemistry reflect the structure and functioning of marine ecosystems and biogeochemical cycles, but they vary in space and time with physical and chemical conditions-and it is this variability that determines the richness of the ocean's responses to environmental change. (A) Marine microbes, primarily phytoplankton in surface layers of the ocean, convert dissolved gases and inorganic nutrients into food. Other microbes break it back down again, governing the cycle of life and death in the sea. By mediating these fundamentally important flows of energy and conversions of matter, marine microbes serve as the foundation of the ocean's food webs and the engines of its biogeochemical cycles. When considered globally and in long-term balance, this cycle of life and death in the sea follows redfield stoichiometry with CO 2 , nitrate, and phosphate being assimilated into organic matter-phytoplankton-and ultimately regenerated in the deep sea with an atomic ratio of close to 106:16:1. in steady state, the upward transport of dissolved inorganic carbon matches the downward flux of carbon in organic matter, so there is no net downward transport of carbon associated with this internal cycling (eppley and Peterson, 1979). The cycle maintains the biologically mediated gradient of CO 2 that was established by depletion of nutrients in the surface layer and decomposition of the resultant organic matter at depth. environmental variability ensures that steady state does not exist, however, and the resultant changes in nutrient distributions and departures from redfield have profound influences on the nitrogen inventory of the ocean, on carbon fluxes, and on food web interactions in the sea. (B) Aquatic systems show remarkable constancy in size spectra, the abundance or amount of biomass in successive logarithmic size bins. This spectrum describes the size distribution of the abundance of planktonic organisms ranging from bacteria (i), to small phytoplankton (ii), to large phytoplankton (iii), and to mesozooplankton (iv), compiled from oceanic and nearshore waters, reservoirs, and high mountain lakes (from Figure 5 in rodríguez, 1994). The relationship has a slope that closely approaches the value of -1, which is consistent with theory proposed by Sheldon et al. (1972) and subsequently studied by many others.
(C) Assuming that energy and biomass flow to larger organisms from smaller ones, the steepness of the slope between size bin (w) and the next larger (w+1) in this biomass spectrum is determined by its biomass, B(w), divided by its turnover time, τ(w). This consumption-related flux between compartments reflects losses due to incomplete assimilation of ingested material and respiration. A more steeply negative slope can imply less-efficient energy flow to higher trophic levels. A bulge in the spectrum implies an accumulation of energy/biomass following a pulse in production (e.g., a spring bloom). This bulge can travel to larger size classes via consumer links, and is attenuated/ diffused according to the properties of each consumer size class. From recent work of author Levin and Charles A. Stock and Thomas M. Powell, both University of California, Berkeley; paper submitted (d) Margalef's mandala (redrawn from Margalef et al., 1979) by Cullen (in press) describes how the community structure of phytoplankton is characterized by functional morphology as it relates to turbulence and nutrients-environmental factors that have a very strong influence on survival of phytoplankton and that are expected to vary with global warming. This model provides no quantitative predictions, but it is generally consistent with observation, experimentation, and theory (Kiørboe, 1993), so it is useful as a guide in predictive modeling.
of the marine N cycle, modulation of N 2 fixation in oligotrophic gyres by supplies of iron from atmospheric deposition, or, conceivably, intentional fertilization.   Arrigo, 2005). The example is for the contemporary ocean, which has a mean N:P slightly less than 16. Waters enriched by upwelling or mixing support high rates of production of organic matter with relatively low N:P. Sinking of this matter and its subsequent degradation leads to oxygen depletion, loss of fixed nitrogen from the ocean through denitrification and anaerobic ammonium oxidation (anammox) reactions, with further reduction of deep ocean N:P. These processes are countered in oligotrophic waters, where stratification and nitrogen depletion favor N 2 fixation and the production of organic matter with high N:P. The trends in the stoichiometry of primary production are consistent with analyses of microbial growth and chemical composition in an evolutionary framework (Klausmeier et al., 2004a(Klausmeier et al., , 2004b, reinforcing the view that the ecology and evolution of marine microbes are integral to biogeochemical cycling in the sea. Reprinted by permission from Macmillan Publishers Ltd: Nature, (Arrigo, 2005)  The year-to-year change in the annual average value of an ocean variable is almost always much smaller than the within-year seasonal variability. This is true of temperature, and also true of microbial abundance. This means that multiyear trends cannot be predicted solely from knowing how microbes respond to short-term environmental change. From our case study, within any year, weekly picoeukaryote abundance is strongly correlated with weekly temperature ( Figure 4A), but not so with weekly nitrate concentration ( Figure 4B).

However, these relationships do not
predict what happens over many years.
The multiyear pattern is only seen when weekly observations are averaged over the whole year. On an annual basis, by and large, picoeukaryotes increase and decrease coherently with nitrate ( Figure 4D), but not with temperature ( Figure 4C).   Smith et al. (2000) describe the bacterial genus Neisseria (which includes the causative agent of gonorrhea) as a commonweatlh, and we indeed envision the relationships between bacterial assemblages that might or might not be called "species" as commonwealth-like. in this figure, similarly colored circles represent bacterial genomes that are sufficiently similar in dNA sequence that recombination (exchange of genes or gene segments) can readily occur between them (indicated by arrows). Circles with multiple colors represent mosaic genomes, different parts of which can exchange with similarly colored regions of other genomes. The enclosed regions A-F are populations, consisting of individuals that interact with each other most frequently, ecologically and genetically. The blue and black boxes represent genes laterally transferred from more remote genomes (outside the "commonwealth"). Such genes may often convey abilities, such as growth on a novel substrate or resistance to antibiotics, that define a population's ecological niche. Because d and F have the same laterally transferred genes, they might exhibit overlapping niches, and yet be unable to exchange information by recombination, while e and F, though recombining, occupy different niches. in general, within-population gene exchange is more frequent than between-population exchange, but there is no reason not to expect that all possible combinations and extents of within-and between-population exchange and degrees of ecological separation will be observed among the ocean's microbes. The quest for some uniform species concept can only confuse our understanding of such situations.
"genomic islands"   Rusch et al., 2007) show that we have consistently underestimated diversity, however this might be defined or measured.
Excitingly, we are starting to have a good idea about how at least some of the apparently rampant oceanic gene swapping occurs: via viruses (Millard et al., 2004;Sullivan et al., 2006). Phages (bac- Rhodopsin genes have, for instance, been transferred on several occasions back and forth between halophilic archaea and bacteria, and among different phyla of the latter, in aid of light-driven proton or chloride pumping and phototaxis (Sharma et al., 2006).  Bouman et al., 2006;Johnson et al., 2006), the stage will be set to describe the interactions of these organisms with each other and with the physical and chemical environment. But the descriptions will be necessarily complex, and it will be difficult to determine, a priori, how much complexity is needed to describe key drivers of the structure and functioning of ecosystems for the scales being studied and the questions being asked.
Recently, Follows et al. (2007) described Interdisciplinary efforts to understand the "sea of microbes" will lead to vastly improved predictions of the ocean's role in a changing world.

ACKNOWled geMeNtS
We thank the editors and the Gordon