The Role of Ecological Theory in Microbial Ecology

Home , Microbial ecology, On the Origin of Species, Productivity (ecology)

PERSPECTIVES

many areas in which microorganisms are of ESSAY environmental and economic importance. For example, improved quantitative theory The role of ecological theory could increase the efficiency of wastewater treatment processes, through the prediction of optimal operating conditions and in microbial ecology conditions that are likely to result in system failure. Quantitative information on the James I. Prosser, Brendan J. M. Bohannan, Tom P. Curtis, Richard J. Ellis, links between microbial community structure, Mary K. Firestone, Rob P. Freckleton, Jessica L. Green, Laura E. Green, population dynamics and activities will also Ken Killham, Jack J. Lennon, A. Mark Osborn, Martin Solan, facilitate assessment and, potentially, mitiga- Christopher J. van der Gast and J. Peter W. Young tion of microbial contributions to climate change, and should lead to quantitative Abstract | Microbial ecology is currently undergoing a revolution, with predictions of the impact of climate change repercussions spreading throughout microbiology, ecology and ecosystem science. on microbial contributions to specific eco- The rapid accumulation of molecular data is uncovering vast diversity, abundant system processes. Given the high abundance, biomass, diversity and global activities of uncultivated microbial groups and novel microbial functions. This accumulation of microorganisms, the ecological theory that data requires the application of theory to provide organization, structure, has been developed for plants and animals mechanistic insight and, ultimately, predictive power that is of practical value, but is of limited value if it does not apply to the application of theory in microbial ecology is currently very limited. Here we microbial communities. Microorganisms argue that the full potential of the ongoing revolution will not be realized if arguably provide much better controlled and more manipulable experimental systems research is not directed and driven by theory, and that the generality of established for testing ecological theory than plants ecological theory must be tested using microbial systems. or animals, and such testing is essential to establish the generality of theory. The use of Bacteria and Archaea have an essential microbial processes (in, for example, better controlled microbial systems might role in earth system processes. They are wastewater treatment, industrial chemical also generate new theory that is relevant to ubiquitous, possess enormous metabolic and production, pharmaceutical production and plants and animals. physiological versatility and are essential to bioremediation), and the realization that Two factors limit the development of virtually all biogeochemical cycling proc- many nonspecific microbial processes such theory in microbial ecology. The first esses — microbial carbon and nitrogen are as biogeochemical cycling are essential for is a lack of data and associated insights. calculated to be, respectively, equivalent to ecosystem sustainability, understanding the This is due in large part to the difficulties and tenfold as great as the carbon and nitro- factors that control these processes is crucial. inherent in observing microorganisms in gen stored in plants1. Although small (~10–6 In our view, this can best be achieved by nature, which often have few distinguishing m), they are abundant (>1030 individuals generating theory that is based on existing morphological features and often cannot globally). Their phylogenetic and physiologi- observations and subsequent experimental be cultivated in the laboratory. The applica- cal diversity is considerably greater than that validation. tion of cultivation-independent molecular of animals and plants and their interactions techniques and their successors — genom- with other life forms are correspondingly The importance of theory ics, metagenomics, transcriptomics and more complex. Theory is used to classify, interpret and proteomics — has generated a plethora of Understanding the ecology of micro- predict the world around us. Without it, new and more comprehensive observations organisms is arguably one of the most microbial ecology is merely the accumula- of microorganisms in nature, but we still compelling intellectual challenges facing tion of situation-bound statements that lack the theoretical tools required to detect contemporary ecology. Although worthy for are of limited predictive ability, providing underlying principles and mechanisms. its intellectual merits alone, developing such microbiologists with few insights. Theory The second factor is cultural, in that the an understanding is essential to meet many has an essential role in developing an under- tools and disciplines of ecological theory of the major challenges facing human society standing of, and explaining the interactions are not part of the contemporary mindset today, such as the management of natural between, microorganisms and their physical, in microbiology. Ecological theory and ecosystems and the mitigation of climate chemical and biological environments. quantitative reasoning typically form only change. Despite this, the application of theory This understanding will be lacking if it is minor components of education in microbi- is severely lacking in microbial ecology solely qualitative, and a full understanding ology, and microbiologists have traditionally where, paradoxically, it is required most. Just therefore requires quantitative theory. used a detailed, reductionist approach that as ecological theory arose from natural his- Theory generates predictions that can be is based on understanding physiological tory to draw generalized conclusions from of practical value for policy makers, stake- mechanisms, with relatively little attention specific observations of organisms in their holders and society. A striking example is the paid to theory. Although the challenge environment, so microbiologists need theory use of epidemiological models to predict for the microbial ecologist might appear to interpret the plethora of observations the spread of human and plant pathogens to be the discovery (or recollection) of ever- that have been made since van Leeuwenhoek and the use of these predictions to inform and more fascinating details of a given system, first saw ‘animalcules’ more than 300 years implement control policies2. There is similar the theoretician aims to predict as much as ago. With the increasing reliance on specific potential value in applying theory in the possible about a system using as few of these

PERSPECTIVES

details as possible; but the populations and systems and allow the much more effective might not lead to simple mapping between structures of microbial communities, by management of the natural world. molecular markers and an ecological niche. comparison with those of plants and ani- In the following sections we discuss More crucially, speciation, and ecological mals, remain inscrutable. The application of examples of areas of ecological theory that species definitions, must consider bacterial molecular techniques has demonstrated the might be particularly valuable in microbial gene-transfer processes, which are erratic need for discovery research, but in our view ecology. In doing this, we attempt to deter- and transfer only a small part of the genome. this can only be exploited if it is directed mine whether the particular characteristics They provide a potential mechanism for by insights gained from the application of of microorganisms present difficulties in maintaining biological species in Mayr’s theory. applying ecological theory that has been sense8, because an incoming gene can developed for plants and higher animals. replace the homologous copy in the genome, Current ecological theory We consider whether and where new theory maintaining the genetic cohesion of the An established body of theory exists for might be required for microorganisms to species. In addition, these processes can also plant and animal ecology but the differ- enhance or replace established ecological result in the horizontal transfer of genes with ences between microorganisms and ‘large’ theory. We also identify conceptual and no counterpart in the recipient that can be organisms, and the extent to which these dif- practical challenges faced by microbial maintained on a plasmid or integrated by ferences restrict the applicability of existing ecologists in applying quantitative ecological non-homologous recombination. However, theory to microbial ecology, often form an theory. the importance of homologous recombina- impasse that is tacitly accepted and seldom tion and horizontal transfer varies widely questioned. Commonly cited differences Ecological species concepts among well-studied bacterial species, and include the small size of microorganisms, Most ecological theory depends on a con- perhaps even more so among the uncultured high rates of population growth, high rates cept of species: population ecology counts masses in the environment. This hetero- and extent of dispersal, the vast abundance individuals within species whereas com- geneity is one reason why we are still far of microorganisms, and the unique aspects of munity ecology and macroecology count from a consensus on the nature of bacterial their biology (such as parasexuality or the number of species. Species are most species, as revealed at a recent Royal Society extremely hardy resting stages). However, commonly defined through the biological discussion meeting9. the breadth of distribution of many of these species concept promoted by Mayr4. This is A consequence of gene transfer is that traits among microorganisms in nature is a genetic definition that envisages a species the bacterial genome is thought to consist not known. Furthermore, the existence of as a group of interbreeding individuals that of two distinct parts, the core genome and these traits does not necessarily prevent the is isolated from other such groups by bar- the accessory genome10. The core genome application of existing ecological theory riers to recombination. If genetic exchange comprises genes that are essential in most to microorganisms (see later discussion of within a species is sufficiently extensive, circumstances and might form the basis for spores and seed banks). Also, the relatively and that between species is sufficiently low, Mayrian species that maintain coherence large scales of time and space over which species will be relatively homogeneous in through homologous recombination. The most microorganisms are studied does themselves and ecologically distinct from accessory genome encodes special ecological not necessarily preclude the application of other species. Unfortunately, prokaryotes adaptations in genes that are readily gained existing theory; theory related to the subdis- (and some eukaryotes) are asexual, thereby and lost. Strains that belong to the same cipline of ecology called macroecology was violating these assumptions, and do not species, as defined by their core genome, can developed specifically to further the under- form species in this genetic way. An differ in the presence and absence of hun- standing of ecology on large scales of space alternative, the ecological species concept, dreds of accessory genes, and consequently and time (see below). The challenge facing defines a species as a set of individuals can have different ecological capabilities. microbial ecologists, and indeed all ecolo- that can be considered to be identical in all According to this view, Cohan’s ecotypes are gists, is to match the appropriate theoretical relevant ecological properties. Cohan5 has merely temporary lineages with particular approach to the organism, system, scale and argued that bacteria have ecological species constellations of accessory genes, and the question of interest. (‘ecotypes’). He postulates that bacteria ecological niche cannot explain the apparent Microbial model systems have played an occupy discrete niches and that periodic cohesion of species that are defined by the important, although often underappreci- selection will purge genetic variation phylogeny of core genes11. ated, part in the development of existing within each niche without preventing Surveys of 16S ribosomal RNA (rRNA) ecological theory (reviewed in REF. 3), divergence between the inhabitants of dif- gene sequences have demonstrated the huge demonstrating its general applicability to ferent niches. So, genetically and ecologi- diversity of bacterial communities, but if microorganisms. However, it is less common cally distinct species will arise, provided much of the interesting ecological adapta- for existing theory to be applied to micro- there is little or no recombination, and tion is conferred by the accessory genome organisms in nature despite the fact that this ecological theories that assume such spe- then the true ecological diversity exists in the would be valuable. It would be extraordinar- cies should apply to bacteria. This also rich brew of catabolic plasmids, resistance ily inefficient to attempt to reinvent existing predicts that molecular diversity should transposons and pathogenesis islands. These theory for application to microorganisms. relate directly to ecological diversity. can be shared among disparate bacteria in an Furthermore, the application of existing Cohan’s ecotypes depend on discrete environment that favours them, but can be theory would afford ecologists the opportu- niches but speciation is more difficult to absent in the ‘same’ bacterial species growing nity to test the true generality of ecological envisage when the relevant environmental elsewhere. The methods of evolutionary principles and to create a synthetic ecology variables are continuous. Bacterial speciation ecology have been applied to the interaction that spans all organisms. This would greatly in these situations could be explored using between these accessory elements and their increase our understanding of ecological the theory of adaptive dynamics6,7, but this host bacteria. For example, Bergstrom and

PERSPECTIVES

hybridization (BOX 1). Gans and colleagues17 Box 1 | Theoretical approaches for estimating diversity in a sample highlight the requirement for collector’s A species abundance curve is simply a graph in a Species curve curves of in excess of one million PCR- which the abundance of a particular species is Area under species curve = St derived clones to ensure coverage of 80% of plotted on the x axis and the number of species at bacterial species within a 1-g soil sample. that abundance is plotted on the y axis (see Without screening large numbers of clones, figure). The observation and contemplation of 1/a sampling low-abundance species remains a these distributions is supported by a rich literature matter of chance. in conventional ecology in which some research, but not all, has imbued such distributions with 16S rRNA gene sequences provide an Log (N ) Log (N ) Log (N ) some ecological meaning. However, microbial 2 min 2 0 2 max operational measure of species. High- 18 ecologists have an interest in species abundance Number of species S throughput sequencing or SARST (serial curves because the area underneath a species analysis of ribosomal sequence tags)19 are abundance curve is the total diversity. This currently the best suited techniques for presents us with a ‘catch 22’ situation: we cannot 0 5 1015 20 estimating prokaryotic diversity. However, measure abundances, so do not know the species– Log2 (bacterial abundance) strains or isolates with identical 16S rRNA area curve, so we cannot estimate diversity. b Individuals curve gene sequences can have different physiolog- In the absence of data, we can assume that a ical characteristics of ecological importance N particular distribution, for example a log-normal max and methods with greater taxonomic resolu- distribution, applies. We can then make an tion are therefore required. Approaches estimate on that basis. Guessing distributions is such as pyrosequencing20,21, which address not a wholly satisfactory procedure. Consequently, diversity across entire metagenomes, might others have sought to fit a line to, and extrapolate Area under individuals curve = Nt from, abundance data (typically clone libraries) be appropriate and could suggest alterna- available to them17. tive conceptual approaches to diversity. Unfortunately, clone libraries in microbial Many ecological questions require infor- ecology are so small (<103) and microbial Number of individuals mation on specific phylogenetic groups communities so large (>1015) that the sample or functional groups, such as rhizobia or distribution is unlikely to look like the ammonia oxidizers, which might increase 0 5 1015 20 community from which it was drawn. An tractability. Log2 (bacterial abundance) alternative approach to estimating species Many of the key questions in microbial abundance curves is to examine the community ecology require reliable estima- 17 reassociation kinetics of DNA extracted from an environment . This approach involves tion of species richness. Analysis of species denaturing DNA, separating the two strands of the DNA molecule, and then allowing them to abundance curves and the lack of a universal reassociate. The most abundant sequences should reassociate first and the reassociation kinetics therefore reflect the underlying distribution of similar sequences and, consequently, definition of species highlight the practical the genomic diversity. However, for experimental reasons, only the reassociation of a small and conceptual difficulties associated with proportion of the diversity can be observed. Consequently, the bulk of the curve is extrapolated such estimates. The analyses described from a few taxa. It can be plausibly argued that this means that there is a great deal of above provide the basis for quantifying uncertainty about the unobserved portion of the species abundance curve. species richness and for assessing the cost The figure shows a log-normal species abundance curve and corresponding cumulative individuals and feasibility of quantification.

curve. 1/a is the width of the species curve, where a is the spread parameter. Nmin is the abundance of

the least abundant species; Nmax is the abundance of the most abundant species; and N0 is the modal Spatial scale species abundance. Figure reproduced with permission from REF. 16 © (2002) US National Academy The pivotal role of spatial patterns and of Sciences. processes in ecology is widely recognized. Many systems, such as fragmented habitats colleagues12 discussed the conditions for describe microbial diversity within any and populations, cannot be studied without plasmid maintenance, and a recent theoreti- given environment. The sheer complex- a serious consideration of space. This has cal exploration concludes that the ‘evolu- ity of most environments, and the rapid generated the subdiscipline of landscape tionary arms race’ between bacteria and realization that collector’s curves of cloned ecology (which has recently been applied bacteriophages can result in speciation of environmental 16S rRNA gene sequences to ecological aspects of antibiotic resistance the host13. This presents a major challenge to would give complete coverage only in the in bacteria22), the metapopulation paradigm those studying prokaryotic population and very simplest ecosystems, has necessitated and metacommunity theory23,24. Other community ecology. The current solution is the development of a more theoretical basis research areas focus directly on the role to use operational definitions of taxonomic for estimating prokaryotic diversity. To this of spatial scaling in ecological patterns. units but we are a long way from a coherent end, Dunbar and colleagues15 and Curtis For example, species–area relationships body of theory that relates the fluid nature of and colleagues16 pioneered the use of species (SARs) have a long history in ecology (see bacterial genomes to the ecology of bacterial abundance curves that use log-normal rela- for example REFS 25–27). A SAR describing communities. tionships (which will include some taxa that areas with relatively few species in com- are rare and others that are present in high mon has greater species turnover28 and is Measuring diversity and species richness numbers) to provide theoretical estimates steeper than a SAR with more species in Since the estimation of substantial micro- of prokaryotic diversity, yielding diversity common; steepness therefore describes bial diversity within soils14, microbial estimates that are similar to those derived by how quickly local assemblages of species ecologists have yearned to quantify and Torsvik and colleagues14 using DNA–DNA differentiate in space.

PERSPECTIVES

The spatial scaling of microbial diversity increased and decreased taxonomic diver- in some instances resemble that of plants is now being addressed by coupling the sity of bacteria in aquatic mesocosms and and animals. Further studies are required molecular characterization of microbial that the shape of the relationship between to determine the mechanisms underlying communities with macroecological theory29. productivity and diversity differed between microbial species richness and the influence Compared with plants and animals, few SAR bacterial taxa. These initial results suggest of nutrient supply (for example, that associ- studies have been published for microorgan- that bacterial diversity can vary with energy ated with eutrophication) on microbial species isms, making a balanced comparison of and that the nature of the relationship can richness and diversity. SAR patterns between the different groups difficult. The SAR is commonly assumed Box 2 | Theoretical approaches for estimating microbial species–area relationships to follow a power-law of the form S ∝ Az, where S is species richness, A is area and z Contiguous habitats Islands is the slope of the curve. Empirical evidence suggests that for animals and plants within 0.3 contiguous habitats, z is generally in the range of 0.1 to 0.2, and for discrete islands z is steeper (0.2 < z < 0.39) (REF. 27), although 0.2

a new meta-analysis of SAR slopes sug- ) gests that this difference might not be as z pronounced as previously thought30 (BOX 2). Slope ( This study also confirmed a general trend in the increasing steepness of z with increasing 0.1 body size from ciliates to large mammals. Recent research has documented power-law species–area (or more generally, taxa–area) 31,32 0.0 relationships in fungi and bacteria k 29,33–36 tan ls and bacteria in ‘island’ habitats . The p um plants z values estimated in studies of contiguous Saltmarsh Marine Arid soil Marine Animals Lake Bioreactor Treehole S Anima habitats were much lower than those of bacteria diatoms fungi ciliates and plants bacteria bacteria bacteria bacteria and island habitats, but island z values were simi- Despite the theoretical and practical importance of species–area relationships (SARs), which lar in magnitude to those observed for plants relate an area (A) to the number of species (S) found within this area, they are difficult to quantify and animals. More research is required directly at ecologically relevant scales (see figure for a comparison of some microbial SARs from to establish whether microorganisms are different contiguous habitats and islands). For organisms with the extraordinary abundance and distributed spatially in ways that are similar diversity of microorganisms, this poses a challenge even at the scale of a single environmental to plant and animal species, but one study sample. Microbial ecologists (and plant and animal ecologists) must therefore use theoretical indicates that soil community composition is approaches to estimate SARs. non-random at a continental scale, and that The most straightforward analyses of microbial SARs are direct plots of sample data (see for soil community composition and diversity example REF. 35). These analyses assume that the slope of the observed sample SAR parallels the slope that would result from a complete census. For a power-law SAR (in which the number of at large scales can be predicted primarily on species is a constant power of the area (S ∝ Az; where z is the slope of the curve)), this translates to 37 the basis of a single variable (pH) . Such an assumption that the observed species richness in a sample is a constant proportion of the total patterns differ from those of plants and species richness in the area from which it was sampled, and that this constant proportion is not animals, the biogeographical distributions affected by scale. of which are influenced by site temperature Parametric approaches are also commonly used to estimate the increase of species richness and latitude. with sample size (or sampling area) (BOX 1). In short, sample data are fitted to models of relative abundance (or assumed on theoretical grounds), and this sample frequency Diversity–energy relationships distribution is projected to estimate the number of unobserved species in the community70. In addition to relationships between diver- Parametric approaches assume that the sample frequency distribution is a truncated version sity and area, common patterns have been of the community-level distribution, which in turn assumes that individuals are randomly described between diversity and energy. sampled from the community. In many studies this assumption can be seriously violated. Microbial communities are commonly investigated by identifying individuals from soil or For example, primary productivity (the sediment cores across a landscape. Even if these environmental samples are randomly rate of energy capture and carbon fixa- distributed in space, spatial aggregation in microbial populations will result in a non-random tion by primary producers) is thought to sample of individuals from the community. be a key determinant of plant and animal An alternative approach to estimating SARs is to examine patterns of community turnover biodiversity 27,38. A positive quadratic or across a landscape (the distance–decay relationship). This method has been applied to estimate hump-shaped relationship is frequently SARs at local, regional and global scales (reviewed in REF. 29). Recent studies have shown that observed between productivity and diversity, distance–decay methods underestimate SAR slopes71, which suggests the need for further in which diversity peaks at intermediate theoretical work in this area. productivity27, although other patterns have The figure shows the slopes of the SARs for contiguous habitat studies of saltmarsh bacteria32, 72 29 72 also been observed39,40. Bacterial communi- marine diatoms , arid soil fungi , and marine ciliates compared with the slopes of the SARs for island habitat studies of lake bacteria36, wastewater treatment bioreactor bacteria34, treehole ties also exhibit such diversity–energy rela- 35 33 41 bacteria and coolant sump tank bacteria . The blue bars show typical values for studies of tionships. Horner-Devine and colleagues animals and plants in these two habitats26. observed that increasing productivity both

PERSPECTIVES

studies of experimental evolution demon- M16E M12B M14C strate that speciation is ‘easy and likely’ for M14F M11 bacteria through niche and environmental M16C partitioning42,43. This might not occur in M12C M12D nature for many microorganisms, for which M14E M14D M15A growth can be limited by access to resources M16A M16B M16D or other aspects of the environment. Mutnovsky Nevertheless, this property of microbial systems provides unique opportunities for M12A 92(86) M14A theoretical ecologists that have not been fully M15B Ren1H1 exploited. M14B Iceland Most larger organisms have a limited distribution range that is due to physical barriers I7 I2 such as lakes, mountain ranges or seas. The chances of allopatric speciation in microorgan- 77(92) isms can be reduced by high abundance and greater dispersal44. A constant influx of 100(99) V10A immigrants into a given habitat would negate V10D the probability of speciation. However, there U3C is increasing evidence of endemicity for some V9A V10B prokaryotic populations (see for example V10C 100(100) REFS 45,46) (FIG. 1). Many pressing questions in Uzon/ Geyser valley U5 Lassen microbial ecology require the consideration of B9D both spatial and temporal scale. Growth rates B7B B9E U3E B10 B6A can vary over several orders of magnitude B8C depending on environmental and nutritional U3D U3B B6B B9F B8B B9B V9B U3A D4 B9C conditions, and speciation will depend on C51 B9A B7A B8A both growth and dispersal. Analysis of the C55A D1 D2 combined effects of these factors on microbial C57A 67(80) community structure, evolution and ecosys- V8 S5 N38C tem function requires quantitative modelling. C58B N16 C57B N38D C58C N36 N38E C58A N23B Variable activity in microorganisms C55B N23C N42 N23A Early molecular studies did not dif- N15B N38A N38G N23D ferentiate between active and inactive N38B 0.001 N38H microorganisms but a large proportion of N15A N38F the cells in a given environment are inactive C64 Yellowstone at any one time47. Several bacterial genera Figure 1 | Evidence for endemicity in prokaryotes. The phylogenetic analysis of sequences of nine form resistant spores, but non-sporulating genetic loci from Sulfolobus strains, isolated from water and sediment samples collected from a nested bacteria can also switch to slow-growing or hierarchy of five geographical locations, is shown. Clades correspond to the five geographical regions, dormant vegetative forms that are resist- showing that strains within a region share a common evolutionary history that is distinct from strains ant to environmental stress48. Inactivity found in other regions. Strains from the same sample are depicted in the same colour. The scale bar in microbial communities will affect © represents 1 substitution per 1,000 sites. Modified with permission from REF. 45 (2003) American many aspects of their ecology, including Association for the Advancement of Science. population dynamics and diversity, and it complicates the application of ecological Temporal scales issues in ecology, such as climate change, in theory, which usually focuses on living and Microorganisms have the potential for rapid which anthropogenic forcing might have an active individuals. Existing theory on the growth and short generation times, relative evolutionary effect on organism–environment role of seed banks in plant ecology might be to those of plants and animals. This potential relationships. When Dykhuizen8 asked ‘Why applicable. The seed bank is important, as is often not realized in natural environments, are there so many species of bacteria?’, he germination will cause temporal variation where nutritional and physicochemical contemplated factors such as low extinction in the observed plant diversity. Dormant conditions can limit growth but, under rates and high speciation. He reasoned seeds have an important role in the succes- favourable conditions, it can lead to varying that bacteria have been able to avoid mass sion of plant communities, but the diversity patterns of microbial diversity over different extinctions over geological time, unlike of the seed bank (potential diversity) and temporal scales. Evolution in microorgan- some larger organisms (such as the dino- that of the established vegetation (real- isms can occur rapidly, particularly under saurs), owing to their ability to withstand ized diversity) often differ considerably49. strong selective pressures, potentially leading rapid changes in environmental conditions. Inactive microbial cells and seeds do not to convergence of ecological and evolution- Unusually high speciation rates would be a contribute directly to ecosystem processes ary timescales. This fundamental property more likely dominant factor in influencing but are important for the resilience of a can be exploited to examine contemporary temporal diversity patterns, and laboratory community to perturbation and might

PERSPECTIVES

REF. 54 become important when environmental ), although the K and Ks concepts are The factors that determine microbial conditions change. Inclusion of inactive not directly equivalent. A further example characteristics such as the timing of cell community members therefore requires the is the use of predator–prey models in epide- division or the allocation of resources to consideration of the nature of the study. miology, which predict cyclical changes in alternative life forms such as spores or other predator and prey abundances. These models persistence states48 can be investigated Competitive strategies form the basis for mathematical approaches using life-history theory, which traditionally Considering all aspects of ecological theory is to understanding the transmission of infec- addresses the trade-offs between growth, dif- beyond the scope of this article, but competi- tious disease55,56, in which the infectious ferentiation and reproduction. Furthermore, tion theory provides an example that has agent is treated as the predator and the host existing theory predicts how organisms already been applied to microbial popula- as the prey, but with modifications describ- combine different life-history traits such tions. Competition theory is particularly well ing, for example, heterogeneous exposure as sexuality and parasexuality62 and the developed and tested for plants50,51. The intro- and behaviour of host and pathogen. combination of different components of duction by Grime50 of a continuum between competitive fitness (for example, growth three competitive strategies of plants provided Behaviour rate, starvation survival or the transition a conceptual breakthrough and could offer a It is often thought that behaviour is a cogni- from low- to high-resource conditions)63 to useful starting point in describing competitive tive process and therefore is not applicable form life-history ‘strategies’. strategies of microbial populations. to microorganisms. However, this is an area Similarly, chemotaxis (the sensing of, In terms of the three CSR strategies in which theory and experimental microbial and movement towards, a higher con- identified by Grime (competitor–stress ecology are closely linked, as changes in centration of a required resource) could tolerator–ruderal), competitors are adapted cellular processes that occur in response to be scrutinized to determine the relative for rapid resource utilization and long-term external signals can be considered ‘behaviour’, benefits of expending energy to seek a more site occupation, stress tolerators are adapted including those that are triggered by envi- favourable patch, using optimal foraging to persist in low-resource environments and ronmental stimuli, such as chemotaxis, spore models with a high concentration of an infe- ruderals are adapted to highly disturbed germination and quorum sensing. Genomics rior resource or a lower concentration of a sites by growing and reproducing quickly. has revealed a vast array of regulatory systems preferred resource. Although the ecological The C and R axes can relate to zymogenous in even the simplest organisms, highlighting role of energy taxis is often discussed64, microorganisms (for example, pseudomonads their ability to process many simultaneous the wealth of existing theory in this area is growing on readily used organic substrates signals57 to optimize performance. Tools largely ignored. The molecular mechanisms released into the rhizosphere), having a that are used by behavioural ecologists, such of chemotaxis in microorganisms are well µ 58 59 high maximum specific growth rate ( max), as game theory , dynamic programming understood, but a theoretical framework persistence and rapid colonization when and optimization theory 60, could be used would undoubtedly improve its under- substrate appears intermittently at high to improve our understanding of microbial standing in an ecological context. One concentrations (for example, animal excreta, behaviour and responses to changes in example is the application of optimal forag- rhizodeposits or leaf litter). The S axis can resources or temperature. The broad applica- ing theory in modelling the bacteriophage relate to autochthonous organisms (such as bility of behavioural theory is highlighted by exploitation of bacterial hosts65. µ cellulolytic bacteria) with low max and using microorganisms to study the evolution The recent application of ecological high substrate affinity (equivalent to a low of sociality 61. game theory (a mathematical approach used

Monod saturation coefficient, Ks), ensuring survival when there is low substrate flow. These analogies are useful, and experimental Box 3 | Areas with potential for the application of ecological theory studies in batch and continuous culture have • Population ecology, including epidemiology. explored the mechanisms that control com- • Interactions of microorganisms with plants, animals and other microorganisms. For example, petition and other microbial interactions in prediction of the influence of microorganisms on the distribution of plants is essential, but terms of growth parameters (see for example current approaches such as climate envelopes ignore the effects on microorganisms. REFS 52,53), but the relationships often break • Community ecology and community assembly. A central question is why so many species can down because of the complexities of life in coexist when there are apparently so few different niches. This problem is potentially more environments such as soil or sediments. tractable for microbial communities for which generation times can be shorter, spatial scale poses These mechanisms include the remarkable less of a problem and diversity is comparable with, if not greater than, that of plants and animals. physiological flexibility of many microor- • Biodiversity–function. The dependency of ecosystems on microbial organisms implies that the ganisms to micro-environmental changes effects of declining microbial diversity will be of significance to the functioning of these systems. and their genetic flexibility (for example, Studying artificial microbial communities, with defined levels of ‘species’ richness and through horizontal gene transfer). measurement of ecosystem function, has been attempted73–75, but the contribution from More direct analogies have been drawn microbial ecologists to biodiversity–function analysis is currently in its infancy. between Monod growth kinetics and logistic • Macroecology. The identification of patterns is particularly relevant to microbial ecology as it is growth of plant and animal populations, frequently difficult to identify individual species or their functions. The influence of climate described in terms of the intrinsic rate of change on the distribution of species is another area of macroecology that has great potential. µ • Biogeochemical cycling. Biogeochemical cycling process models traditionally treat microbial increase (r) (equivalent to max) and the carrying capacity (K). Concepts of selection communities as a black box, ignoring both biomass and diversity. The establishment of that are based on r and K have been used to quantitative links between microbial diversity and ecosystem processes is essential, and requires scaling from the micron level, at which mechanisms controlling diversity can operate, through to describe bacterial growth under different landscape and global levels, at which important effects are considered. substrate-supply conditions (see for example

PERSPECTIVES

to model situations in which two or more theoretical approaches. This restricts the Fear of failure promotes conservative decision organisms interact) to microorganisms synthesis and integration of data that are gen- making, leading to excessive use of resources significantly affects our understanding of erated by the plethora of techniques available, (for example, environmental engineers use positive interactions, such as cooperation, prevents the introduction of new ideas that too much power and medics overprescribe). among microorganisms66,67. In any system in transcend our experience, prevents the identifi- Furthermore, empirical problem solving is which members of a population cooperate cation of, and focus on, fundamental questions, inevitably subject to the law of diminishing there is potential for defection, which can lead novel approaches and critical techniques, and returns: the increments delivered by each item to the coexistence of multiple behavioural restricts our ability to predict. This has implica- of research get smaller and smaller each year. strategies. The coexistence of individuals that tions for the basic science of microbiology and In many cases we might be able to borrow display different strategies can be explored for its practical applications. New technologies and adapt ecological theory that has already readily in microbial populations using game will increasingly lead us down blind, non-gen- been developed for plants and animals. theory models that were generated for similar eralist and expensive alleyways if studies are Conversely, existing ecological theory must phenomena in multicellular organisms68,69. not directed and driven by theory. be tested and novel theory generated in The relevance and importance of micro- microbial systems. Indeed, the greater control Conclusions organisms in natural ecosystems are self- and manipulation provided by microbial The questions that microbiologists are asking evident, but stakeholders and end-users require experimental systems facilitate more rigorous are inherently quantitative and advances in predictive modelling. Practitioners and policy and thorough tests of ecological theory. In theory require a collaboration with other makers must make decisions about microbial other areas, new theoretical and conceptual disciplines, including ecology. Some of the communities and processes, but in the absence approaches might be required to deal with the main areas with potential for the application of a theoretical framework, many decisions rely smaller size, faster growth, greater dispersal of ecological theory are listed in BOX 3. on a combination of empiricism and intuition. and asexuality of microorganisms. It is Microbial ecology is said to be driven This can deliver successes, but solutions are certain, however, that the central and global — but also limited — by techniques. However, often partial and situation-bound and the importance of microorganisms in natural we propose that advances in microbial ecology failure of empirical and intuitive decision ecosystems necessitates the acceptance, devel- are limited by a lack of conceptual and making can be baffling and counterintuitive. opment and application of ecological theory.

Glossary

Adaptive dynamics Macroecology Niche Adaptive dynamics links evolutionary dynamics and The study of broad-scale patterns in ecology. These include The particular set of resources and environmental population dynamics and emphasizes ecological the patterns in, and relationships between, key ecological conditions that an individual species exploits. This includes interactions in describing the evolution of a population. variables (such as population size and population food, shelter and climatic tolerances. Basic ecological occupancy), as well as life-history parameters that are theory predicts that no two species can coexist if they have Allopatric speciation important in determining ecological niches and responses the same niche, unless they are identical in all respects. Speciation that is due to the physical isolation of to ecological changes (for example, body size, metabolic populations by an extrinsic barrier, which results in rate and growth rate). Optimal foraging models (genetic) reproductive isolation of the populations, such Optimal foraging models aim to describe and quantitatively that if the barrier between the populations breaks down Metacommunity predict behaviour decisions by animals to optimize foraging, individuals from each population can no longer Metacommunities are large-scale regional assemblages of particularly with regard to energy intake and expenditure. interbreed. trophically similar individuals and species, each of which is perceived to exist as a series of local communities, linked Population Autochthonous organisms by the dispersal of potentially interacting species. The A group of individuals of one species in an area that is Autochthonous organisms are adapted to a regular dynamics that arise within metacommunities consist of separate from other groups apart from rare migration supply of substrate at low concentration and have a spatial dynamics and community dynamics (multispecies events. In practice, the size and nature of the area is relatively low maximum specific growth rate and high interactions or the emergent properties arising from them defined, often arbitrarily, for the purposes of the study substrate affinity. within communities), and the interaction between them. being undertaken. Climate envelopes Metapopulation Pyrosequencing The range of climatic conditions under which a population A group of populations that are perceived to exist as a A method for DNA sequencing, in which the inorganic of a species can persist. In reality, climate envelopes are an series of local populations that are linked by migration pyrophosphate that is released from a nucleoside abstraction and species distributions are constrained and between them. However, the rate of migration is limited, triphosphate during DNA chain elongation is detected by a explained by a number of factors beyond climate, such as such that the dynamics of the metapopulation should be bioluminometric assay. habitat availability, historical events, dispersal limitation and interspecific interactions. seen as the sum of the dynamics of the individual subpopulations. Specifically, the size of the metapopulation SARST (serial analysis of ribosomal sequence tags) Community is determined by the balance between extinction and A high-throughput method for characterizing microbial Broadly, this is a collection of populations of different colonization. diversity that is based on cloning and sequencing short 16S species that occur together in space and time. The ribosomal RNA gene sequences linked into concatamers. definition of a community varies. One definition includes all Monod growth kinetics species (that is, across all trophic levels). A less inclusive Monod growth kinetics describe the influence of the Seed bank definition includes all trophically similar species (for concentration of a growth-limiting substrate on the specific A store of viable seed buried and dormant within the example, all the plants in a rainforest). growth rate of microorganisms, in a form similar to the environment. Michaelis–Menten equation. Kinetics are determined by Eutrophication two growth parameters: the maximum specific growth rate Zymogenous microorganisms The enrichment of a water body with nutrients, leading to (achieved at high, non-limiting substrate concentration) Zymogenous microorganisms are adapted to growth on an excessive growth of algae and other photosynthetic and the saturation coefficient, which is the substrate intermittent supply of substrate at high concentrations and organisms, the subsequent decay of which results in the concentration at which the specific growth rate is half of have a relatively high maximum specific growth rate and depletion of oxygen. the maximum specific growth rate. low substrate affinity.

PERSPECTIVES

James I. Prosser and Ken Killham are at the School of 11. Wertz, J. E., Goldstone, C., Gordon, D. M. & Riley, M. 37. Fierer, N. & Jackson, R. B. The diversity and Biological Sciences, University of Aberdeen, Aberdeen A. A molecular phylogeny of enteric bacteria and biogeography of soil bacterial communities. AB24 3UU, Scotland, UK. implications for a bacterial species concept. J. Evol. Proc. Natl Acad. Sci. USA 103, 626–631 Biol. 16, 1236–1248 (2003). (2006). Brendan J. M. Bohannan is at the Center for Ecology 12. Bergstrom, C. T., Lipsitch, M. & Levin, B. R. Natural 38. Evans, K. L., Greenwood, J. J. D. & Gaston, K. J. and Evolutionary Biology, University of Oregon, selection, infectious transfer and the existence Dissecting the species-energy relationship. Proc. conditions for bacterial plasmids. Genetics 155, R. Soc. Lond. B 272, 2155–2163 (2005). Eugene, Oregon 97403-5829, USA. 1505–1519 (2000). 39. Abrams, P. A. Monotonic or unimodal diversity– Tom P. Curtis is at the School of Civil Engineering and 13. Weitz, J. S., Hartman, H. & Levin, S. A. Coevolutionary arms races between bacteria and productivity gradients: what does competition Geosciences, Cassie Building, University of Newcastle bacteriophage. Proc. Natl Acad. Sci. USA 102, theory predict? Ecology 76, 2019–2027 upon Tyne NE1 7RU, UK. 9535–9540 (2005). (1995). 14. Torsvik, V., Goksoyr, J. & Daae, F. L. High diversity in 40. Mittelbach, G. G. et al. What is the observed Richard J. Ellis is at the Natural Environment Research DNA of soil bacteria. Appl. Environ. Microbiol. 56, relationship between species richness and Council (NERC) Centre for Population Biology, Imperial 782–787 (1990). productivity? Ecology 82, 2381–2396 (2001). College London, Silwood Park Campus, Ascot, 15. Dunbar, J., Barns, S. M., Ticknor, L. O. & Kuske, C. R. 41. Horner-Devine, M. C., Leibold, M. A., Smith, V. H. & Berkshire SL5 7PY, UK. Empirical and theoretical bacterial diversity in four Bohannan, B. J. M. Bacterial diversity patterns along Arizona soils. Appl. Environ. Microbiol. 68, a gradient of primary productivity. Ecol. Lett. 6, Mary K. Firestone is at the Department of 3035–3045 (2002). 613–622 (2003). 16. Curtis, T. P., Sloan, W. T. & Scannell, J. W. Estimating Environmental Science, Policy and Management, 42. Helling, R. B., Vargas, C. N. & Adams, J. Evolution of prokaryotic diversity and its limits. Proc. Natl Acad. 151 Hilgard Hall, University of California, Berkeley, Escherichia coli during growth in a constant Sci. USA 99, 10494–10499 (2002). California 94720-3110, USA. 17. Gans, J., Wolinsky, M. & Dunbar, J. Computational environment. Genetics 116, 349–358 (1987). 43. Lenski, R. E., Rose, M. R., Simpson, S. C. & Rob P. Freckleton and A. Mark Osborn are at improvements reveal great bacterial diversity and high toxicity in soil. Science 309, 1387–1390 Tadler, S. C. Long-term experimental evolution in the Department of Animal and Plant Sciences, (2005). Escherichia coli. I. Adaptation and divergence during University of Sheffield, Western Bank, 18. Eckburg, P. B. et al. Diversity of the human 2000 generations. Am. Nat. 138, 1315–1341 Sheffield S10 2TN, UK. intestinal microbial flora. Science 308, 1635–1638 (1991). (2005). 44. Finlay, B. J. & Clarke, K. J. Ubiquitous dispersal Jessica L. Green is at the School of Natural Sciences, 19. Neufeld, J. D., Yu, Z., Lam, W. & Mohn, W. W. Serial of microbial species. Nature 400, 828–828 University of California, Merced, PO Box 2039, analysis of ribosomal sequence tags (SARST): a high- (1999). California 95344, USA. throughput method for profiling complex microbial 45. Whitaker, R. J., Grogan, D. W. & Taylor, J. W. communities. Environ. Microbiol. 6, 131–144 Geographic barriers isolate endemic populations of Laura E. Green is at the Ecology and Epidemiology (2004). hyperthermophilic Archaea. Science 301, 976–978 20. Edwards, R. A. et al. Using pyrosequencing to shed Group, Department of Biological Sciences, University (2003). of Warwick, Coventry CV4 7AL, UK. light on deep mine microbial ecology. BMC Genomics 20 March 2006 (doi:10.1186/1471-216 46. Cho, J. C. & Tiedje, J. M. Biogeography and degree of endemicity of fluorescent Pseudomonas strains in Jack J. Lennon is at the Macaulay Land Use Research 4-7-57). soil. Appl. Environ. Microbiol. 66, 5448–5456 Institute, Craigiebuckler, Aberdeen AB15 8QH, UK. 21. Sogin, M. L. et al. Microbial diversity in the deep sea and the underexplored ‘rare biosphere’. (2000). Martin Solan is at the Oceanlab, University of Proc. Natl Acad. Sci. USA 103, 12115–12120 47. Bakken, L. R. in Modern Soil Microbiology (eds van Elsas, J. D., Trevors, D. J. & Wellington, E. M. H.) Aberdeen, Main Street, Newburgh, Aberdeenshire (2006). 22. Singer, R. S., Ward, M. P. & Maldonado, G. Can 47–61 (Marcel Dekker, New York, 1997). AB41 6AA, UK. landscape ecology untangle the complexity of 48. Balaban, N. Q., Merrin, J., Chait, R., Kowalik, L. & 4 Christopher J. van der Gast is at the NERC Centre for antibiotic resistance? Nature Rev. Microbiol. , Leibler, S. Bacterial persistence as a phenotypic 943–952 (2006). Ecology and Hydrology at Oxford, Mansfield Road, switch. Science 305, 1622–1625 (2004). 23. Hanski, I. Metapopulation Ecology (Oxford Univ. 49. Edwards, G. R. & Crawley, M. J. Herbivores, seed Oxford OX1 3SR, UK. Press, Oxford, UK, 1999). banks and seedling recruitment in mesic grassland. 24. Hubbell, S. P. The Unified Neutral Theory of J. Ecol. 87, 423–435 (1999). J. Peter W. Young is at the Department of Biology 3, Biodiversity and Biogeography (Princeton Univ. 50. Grime, J. P. Plant Strategies, Vegetation Processes, University of York, PO Box 373, York YO10 5YW, UK. Press, Princeton, New Jersey, 2001). 25. Arrhenius, O. Species and area. J. Ecol. 9, 95–99 and Ecosystem Properties (Wiley, Chichester, Correspondence to J.I.P. (1921). 2001). e-mail: [email protected] 26. MacArthur, R. H. & Wilson, E. O. The Theory of 51. Tilman, D. Plant Strategies and the Dynamics and Structure of Plant Communities (Princeton Univ. Press, doi: 10.1038/nrmicro1643 Island Biogeography (Princeton Univ. Press, Princeton, New Jersey, 1967). Princeton, New Jersey, 1988). 1. Whitman, W. B., Coleman, D. C. & Wiebe, W. J. 27. Rosenzweig, M. L. Species Diversity in Space and 52. Harder, W. & Dijkhuizen, L. Strategies of mixed Prokaryotes: the unseen majority. Proc. Natl Acad. Time (Cambridge Univ. Press, Cambridge, UK, substrate utilization in microorganisms. Philos. Trans. Sci. USA 95, 6578–6583 (1998). 1995). R. Soc. Lond. B. 297, 459–480 (1982). 2. Green, L. E. & Medley, G. F. Mathematical modelling 28. Koleff, P., Gaston, K. J. & Lennon, J. J. Measuring 53. Gottschal, J. C. Some reflections on microbial of the foot and mouth disease epidemic of 2001: beta diversity for presence—absence data. J. Anim. competitiveness among heterotrophic bacteria. Ecol. 72, 367–382 (2003). strengths and weaknesses. Res. Vet. Sci. 73, Antonie Van Leeuwenhoek 51, 473–494 29. Green, J. & Bohannan, B. J. M. Spatial scaling of 201–205 (2002). (1985). microbial biodiversity. Trends Ecol. Evol. 21, 3. Jessup, C. M. et al. Big questions, small worlds: 54. Velicer, G. J. & Lenski, R. E. Evolutionary trade-offs microbial model systems in ecology. Trends Ecol. Evol. 501–507 (2006). 30. Drakare, S., Lennon, J. J. & Hillebrand, H. The under conditions of resource abundance and scarcity: 19, 189–197 (2004). imprint of the geographical, evolutionary and experiments with bacteria. Ecology 80, 1168–1179 4. Mayr, E. in The Species Problem (ed. Mayr, E.) 1–22 ecological context on species–area relationships. (1999). (American Association for the Advancement of Ecol. Lett. 9, 215–227 (2006). 55. Anderson, R. M. & May, R. M. The invasion, Science, Washington DC, 1957). 31. Green, J. L. et al. Spatial scaling of microbial persistence and spread of infectious diseases within 5. Cohan, F. M. What are bacterial species? Annu. Rev. eukaryote diversity. Nature 432, 747–750 animal and plant communities. Philos. Trans. R. Soc. Microbiol. 56, 457–487 (2002). (2004). Lond. B. 314, 533–570 (1986). 6. Metz, J. A. J., Geritz, S. A. H., Meszéna, G., Jacobs, F. 32. Horner-Devine, M. C., Carney, K. M. & Bohannan, 56. Krebs, C. J. Ecology: The Experimental Analysis of J. A. & van Heerwaarden, J. S. in Stochastic and B. J. M. An ecological perspective on bacterial Spatial Structures of Dynamical Systems (eds Distribution and Abundance (HarperCollins College biodiversity. Proc. R. Soc. Lond. B 271, 113–122 Publishers, New York, 2001). van Strien, S. J. & Verduyn Lunel, S. M.) 183–231 (2004). 57. Hellingwerf, K. J. Bacterial observations: a (North Holland, Amsterdam, The Netherlands, 1996). 33. van Der Gast, C. J., Lilley, A. K., Ager, D. & rudimentary form of intelligence? Trends Microbiol. 7. Dieckmann, U. & Doebeli, M. On the origin of species Thompson, I. P. Island size and bacterial diversity in by sympatric speciation. Nature 400, 354–357 an archipelago of engineering machines. Environ. 13, 152–158 (2005). (1999). Microbiol. 7, 1220–1226 (2005). 58. Sigmund, K. Games of Life (Oxford Univ. Press, 8. Dykhuizen, D. E. Santa Rosalia revisited: why are there 34. van der Gast, C. J. et al. Bacterial diversity is Oxford, UK, 1993). so many species of bacteria? Antonie van determined by volume in membrane bioreactors. 59. Mangel, M. & Clark, C. W. Dynamic Modeling in Leeuwenhoek 73, 25–33 (1998). Environ. Microbiol. 8, 1048–1055 (2006). Behavioral Ecology (Princeton Univ. Press, Princeton, 9. Fisher, M. C., Spratt, B. G. & Staley, J. T. (eds) Species 35. Bell, T. et al. Larger islands house more bacterial New Jersey, USA, 1989). and speciation in microorganisms. Philos. Trans. taxa. Science 308, 1884 (2005). 60. Sutherland, W. J. The best solution. Nature 435, 569 R. Soc. Lond. B 361, 1897–2053 (2006). 36. Reche, I., Pulido-Villena, E., Morales-Baquero, R. & (2005). 10. Young, J. P. W. et al. The genome of Rhizobium Casamayor, E. O. Does ecosystem size determine 61. West, S. A., Griffin, A. S., Gardner, A. & Diggle, S. P. leguminosarum has recognizable core and accessory aquatic bacterial richness? Ecology 86, 1715–1722 Social evolution theory for microorganisms. Nature components. Genome Biol. 7, R34 (2006). (2005). Rev. Microbiol. 4, 597–607 (2006).

PERSPECTIVES

62. Redfield, R. J., Schrag, M. R. & Dean, A. M. The 68. Brown, S. P. & Johnstone, R. A. Cooperation in the 74. Woodcock, S., Curtis, T. P., Head, I. M., Lunn, M. & evolution of bacterial transformation: sex with dark: Signalling and collective action in quorum- Sloan, W. T. Taxa–area relationships for microbes: the poor relations. Genetics 146, 27–38 sensing bacteria. Philos. Trans. R. Soc. Lond. B 268, unsampled and the unseen. Ecol. Lett. 9, 805–812 (1997). 961–965 (2001). (2006). 63. Vasi, F., Travisano, M. & Lenski, R. E. Long-term 69. Redfield, R. J. Is quorum sensing a side effect of 75. Azovsky, A. I. Size-dependent species–area experimental evolution in Escherichia coli. II. diffusion sensing? Trends Microbiol. 10, 365–370 relationships in benthos: is the world more diverse Changes in life-history traits during adaptation to a (2002). for microbes? Ecography 25, 273–282 seasonal environment. Am. Nat. 144, 432–456 70. Wohl, D. L., Arora, S. & Gladstone, J. R. Functional (2002). (1994). redundancy supports biodiversity and ecosystem 64. Alexandre, G., Greer-Phillips, S. & Zhulin, I. B. function in a closed and constant environment. Ecology Acknowledgements Ecological role of energy taxis in microorganisms. 85, 1534–1540 (2004). This article arose from a workshop supported by UKPopNet, FEMS Microbiol. Rev. 28, 113–126 71. Battin, T. J., Kaplan, L. A., Newbold, J. D. & Hansen, and funding is gratefully acknowledged. (2004). C. M. E. Contributions of microbial biofilms to 65. Bull, J. J., Pfennig, D. W. & Wang, I. N. Genetic details, ecosystem processes in stream mesocosms. Nature Competing interests statement The authors declare no competing financial interests. optimization and phage life histories. Trends Ecol. 426, 439–442 (2003). Evol. 19, 76–82 (2004). 72. Naeem, S., Hahn, D. R. & Schuurman, G. Producer- 66. Velicer, G. J. Social strife in the microbial world. Trends decomposer co-dependency influences biodiversity FURTHER INFORMATION Microbiol. 11, 330–337 (2003). effects. Nature 403, 762–764 (2000). Natural Environment Research Council homepage: 67. Travisano, M. & Velicer, G. J. Strategies of microbial 73. Bunge, J. & Fitzpatrick, M. Estimating the number of http://www.nerc.ac.uk UKPopNet: http://www.ukpopnet.org cheater control. Trends Microbiol. 12, 72–78 species: a review. J. Am. Stat. Assoc. 88, 364–373 Access to this links box is available online. (2004). (1993).