Microbial Landscapes: New Paths to Biofilm Research

PERSPECTIVES

the relationships between seed transport, OPINION vegetation patterns and the landscape. So why not import the tenets of landscape ecology and Microbial landscapes: new paths to related fields into biofilm research? In this article, we propose that biofilms are microbial landscapes, to emphasize their spatially biofilm research explicit dimension, and to lay the foundation for a unifying theoretical basis that supports Tom J. Battin, William T. Sloan, Staffan Kjelleberg, Holger Daims, Ian M. Head, and guides the relationships between biodi- Tom P. Curtis and Leo Eberl versity, ecosystem function and the effects of scales (composition, structure or function) Abstract | It is the best of times for biofilm research. Systems biology approaches are in biofilm research. providing new insights into the genetic regulation of microbial functions, and sophisticated modelling techniques are enabling the prediction of microbial Biofilms are landscapes community structures. Yet it is also clear that there is a need for ecological theory to When we view a landscape, we look at the contribute to our understanding of biofilms. Here, we suggest a concept for biofilm composition and expanse of elevations, valleys, rivers, vegetation and animals. research that is spatially explicit and solidly rooted in ecological theory, which might Landscape ecology essentially studies the serve as a universal approach to the study of the numerous facets of biofilms. interactions between the spatial patterns of these landscape elements and the ecological The importance of microorganisms in behaviour that has led to the perception of processes operating within and between human health and disease, and the massive biofilms as cities of microorganisms5. them14. Landscapes themselves can be impact of the pure-culture approach devised The intense research on single- or multi- defined as the configuration of patterns at by Robert Koch and others, has understand- species biofilms grown in flow cells1–3 and the any scale relative to the ecological processes ably led to a philosophy in microbiologi- continuing attempts to understand the role or organisms under investigation, and the cal research that emphasizes the study of of biofilms in situ9–11 have both progressed concept of landscape can therefore be applied microorganisms in pure liquid culture. This our understanding of biofilm communities. to any scale or system14. The spatial patterns approach has so prominently pervaded Flow-cell work under standardized and opti- of organisms primarily result from abiotic microbiology that biofilm research was long mized laboratory conditions echoes formal factors, such as climate and landform, but neglected until microbiologists ‘re-discovered’ community theory that focuses on a single organisms can also physically alter their envi- these fascinating communities almost 40 scale (in which scale is the temporal measure ronment and create spatial heterogeneity14. years ago1–3. Growing appreciation of the of a pattern or a process, or the level or Resulting patches of resources and organisms importance of biofilms has now led to the degree of spatial resolution) and assumes that are delineated by boundaries that control perception that these communities consti- local communities are closed and isolated12. processes within the patches and regulate the tute the dominant mode of microbial life These laboratory experiments have unrav- flow of organisms and information through in most aquatic ecosystems1–3. Wherever elled many microbial interactions, including the landscape14–16. For instance, animals often free-swimming microbial cells encounter competition and cooperation, which are move from one location to another because surfaces, they can switch from a planktonic largely deterministic in nature and relate to of the patchy distribution of resources lifestyle to form sessile communities that are theories of coexistence by niche differentiation. and, at the same time, the heterogeneity of enclosed by a slimy matrix. These commu- In the wild, however, biofilms are open and landscape elements (for example valleys and nities often form striking architectures. highly dynamic communities and exist as part mountains) can enhance or inhibit their A growing body of excellent reviews3–8 of a larger microbial network. In this network, movement. This has logical implications for has plotted the highlights of biofilm local communities are linked by dispersal dispersal and the colonization of new niches research carried out over the last decade. and multiple interacting species to form a or the invasion of existing communities. Biofilms are now recognized as complex metacommunity. This introduces an explicitly Patches and their boundaries also affect the and dynamic communities in which sub- spatial dimension to biofilm research, and, at transmission of communication signals, stantial phenotypic diversification allows the same time, calls for biofilm research that is as canopy foliage, for instance, can absorb, microorganisms to adapt to different envi- solidly rooted in ecological concepts and uni- attenuate and filter the song of birds. Yet ronments. For instance, the hydrodynamic fying principles. Yet a sober examination of boundaries in a patchy landscape also control conditions and the availability of substrate the latest research reveals that microbiologists material and energy fluxes and thereby and nutrients can shape biofilm architec- lack such a framework13. Undoubtedly this ecosystem functioning. Landscape ecology ture, whereas certain genes are essential for is largely due to the diversity of the biofilm therefore integrates various fields, such as regulating the production of extracellular inhabitants and the habitats they occupy. The neutral ecology, dispersal ecology, invasion polymeric substances. Cell–cell communica- absence of a theoretical framework severely ecology and ecosystem ecology, and is at the tion can control biofilm development and restricts the extension of biofilm research forefront of ecological science16. architecture and, along with programmed across multiple scales and limits our under- We advocate that biofilms should be cell death, seems to drive coordinated standing of biofilms and their community viewed as microbial landscapes (FIG. 1a) and, differentiation in mature biofilms and the interactions at the systems level. at the same time, that they are intercon- release of dispersal cells. Microbiologists We have all walked through woods and nected parts of the larger landscape that they have discovered an unexpectedly high seen seeds that are dispersed by the wind colonize. For instance, a biofilm in a stream degree of coordinated multicellular and carried away — and intuitively we accept is a system, but this system also interacts

PERSPECTIVES

with the surrounding landscape that is biofilms and locally reduce biomass, which is Whether trees grow protected on the formed by the streambed and the catchment. comparable to the wind-fall of canopy trees valley bottom or are exposed to wind on a Similarly, biofilms that form in the lungs of and the resulting gaps in forests. Grazing by mountain, it is the basic landscape topog- individuals with cystic fibrosis3 are integrated protozoa can further contribute to the patchy raphy and the related aerodynamics that into the lung landscape, and are exposed to distribution of biomass17 (FIG. 1a.). More gen- shape their canopies. Similarly, the texture the spatial heterogeneity of hydration and erally, boundary conditions induced either of the colonized surface determines the respiratory mucous viscosity. by wind or water dynamics are analogous in primary structure of biofilm topography. What do forested and microbial land- diverse situations including: the canopy of Whether biofilms coat a flat or a rough scapes have in common? In both landscapes, forests18, mosses19, soil crusts20 and biofilms21. surface determines their exposure to shear multiple physical and biological factors shape Such boundaries not only determine gaseous stress and mass transfer — and these sources the spatial configuration of the biomass. For and dissolved mass transfer, but also influence of spatial variation are rarely encountered in instance, flow-induced disturbance can erode the transport of seeds and other propagules. flow cells. In addition to the resident biofilm

a Figure 1 | Biofilms as microbial landscapes. a | Bacterial biofilms colonizing a lung epithe- lium (left) that could also represent a single- or multispecies biofilm grown in a flow cell, and a microbial biofilm colonizing a sedimen- tary environment in the ‘wild‘ (right). Biofilms coalesce and accumulate biomass in slow flow 2 (1), but predominantly develop filamentous streamers floating in the fast flow (2). Turbulent flow and eddies that are induced by 3 the landscape topography might be preferen- 4 tial trajectories for dispersal cells to land (3). Grazing by protozoa increases the spatial 1 heterogeneity of the microbial landscape (4). b | Neutral theory predicts that demographic stochasticity (death, reproduction) but also immigration from a source community (meta- b Source community community) shapes local community assembly. Biotic interactions and cell–cell signalling might function as filters (dashed line) that dif- c ferentially influence the invasion success of immigrating cells. c | Confocal micrograph of microcolonies of a Pseudomonas aeruginosa biofilm with localized regions where cell death occurs. Green fluorescent cells are viable, whereas red fluorescent cells have a compromised cell membrane and are dead. d | Dispersal behaviour of microorganisms as shaped by the interplay between biofilm Local community topography and induced flow fields around d landscape structures. The dispersal kernel shows the probability of encountering propagules as a function of the distance from their source community. The flow field around mushroom-like structures forces propagules e to the surface, which narrows the dispersal kernel. Microbial landscapes with a less heterogeneous topography induce different flow

Dispersal probability Dispersal fields and larger recruitment areas for propagules. e | Confocal image showing the locali- Distance from parental population zation of nitrifying bacteria in a wastewater biofilm visualized with fluorescence in situ hybridization (FISH) with probes targeting sublineage I Nitrospira (red), sublineage II Nitrospira (green) and ammonia oxidizing bacteria (blue). Part (c) is reproduced with permission from REF. 36 © (2003) American Society of Microbiology. Part (e) is reproduced Dispersal probability Dispersal with permission from REF. 45 © (2006) Distance from parental population Blackwell Science.

PERSPECTIVES

organisms (bacteria and archaea) various Neutral theory and microbial landscapes cal descriptions of biogeography to micro- architects, including algae, protozoa and It has typically been assumed that envi- bial communities. Yet the advent of neutral even small metazoa, construct and reorgan- ronmental factors alone shape microbial theory, itself rooted in biogeography, offers ize the biofilm matrix and confer a further biodiversity, an assumption that until recently microbial ecologists the opportunity to test level of structural heterogeneity to biofilms. also pervaded thinking in the ecology of large the importance of dispersal and chance in This level is characterized by the channel organisms, but has been challenged by the shaping microbial communities29–31. We are networks and protuberances, such as mush- neutral theories of Hubbell24 and Bell25 (BOX 1). convinced that a landscape approach will room-like caps and filamentous streamers. Neutral theory assumes that all individuals in further foster this development. Our perception of biofilms as landscapes a community are strictly equivalent regarding The hypothesis that the intrinsic stochastic has several important implications for reproduction and death, and that community demography of a biofilm affects its structure biofilm research. Biofilms that form part of a assembly is governed by the balance between would seem to be a logical and attractive start- landscape are subject to the same basic rules the dynamics of these demographic processes ing point in the exploration of biofilm land- that couple ecological patterns and processes and random dispersal from a metacommu- scapes. However, this approach is not trivial at larger scales. This concept provides the nity. Neutral theory is therefore a stochastic given the spatial heterogeneity that is inherent framework to place biofilms in the context of theory, and given that dispersal from a source to biofilms. Nevertheless, Houchmandzadeh microbial biogeography, the discipline that to a local community (FIG. 1b) is the main and Vallade32 showed that stochastic studies the large-scale distribution and move- mechanism controlling community assembly, processes alone can shape the spatial distribu- ment of microorganisms22,23. This framework it also is a spatial theory26. tion of taxa with self-similarities at all scales also paves the way for a spatial theory of the Although theories on the spatial patterns and biomass clustering, as observed in real biodiversity of biofilms, and enables connec- formed in microbial communities have communities. Strikingly, their simulated tions to neutral theory and the related fields been proposed27,28, microbial ecologists have landscapes are extremely similar to the of dispersal and invasion ecology. been slow to apply quantitative mathemati- spatial heterogeneity of stream biofilms (see Supplementary information S1 (figure)). Although new technologies are elucidating Glossary patterns in the topography of biofilms, we currently cannot explicitly observe biodi- Commensalism Mutualism Commensalism is an interaction between two organisms in Mutualism is an interaction between two or more species versity patterns at the scale of ‘landscape which one organism benefits and the other is neither in which both species derive benefit elements’ in biofilms. Unlike classical ecology, harmed nor helped. in which community patterns are easy to Neutral ecology observe and in which models can give invalu- Dispersal ecology Neutral ecology aims to explain the diversity and relative Dispersal ecology studies the processes by which a species abundance of species in ecological communities. It assumes able insights into the underlying mechanisms maintains or expands the distribution of its populations. that the differences between members of an ecological that drive community composition, biodi- Dispersal implies movement away from an existing community of trophically similar species are ‘neutral,’ or versity patterns in biofilms are more difficult population (population expansion) or away from the parent irrelevant to their success. to assess. Mathematical descriptions of organisms (population maintenance). Niche differentiation ecological patterns need to be parameterized Dispersal kernel Niche differentiation refers to the process by which natural by observations at small scales, and only by The spatial distribution of offspring around a parent selection drives competing species into different uses of integrating or ‘up-scaling’ them, will land- can be described as a function of the probability of resources resulting in different niches. scape descriptions emerge. the abundance of offspring and distance from the parental community. This function is represented by Rhizosphere Neutral community models are surpris- the kernel model. The rhizosphere is the narrow region of soil that is directly ingly good at predicting the broad-scale influenced by roots and associated soil microorganisms. patterns that are observed in microbial Extracellular polymeric substance Abundant microorganisms feed on sloughed-off plant cells, communities29–31, but it is likely that neutral Extracellular polymeric substances are the key and the proteins and sugars released by roots. components of the slime matrix of biofilms, and are community assembly forms just one part of composed of polysaccharides, proteins, nucleic acids, Self-similarity a more complex system in which the pat- lipids and other macromolecules. A self-similar object is exactly or approximately similar to a terns from deterministic processes override part of itself. those imposed by neutral processes. In fact, Landscape ecology 26 Landscape ecology addresses the causes and Shear stress as pointed out by Alonso and colleagues , consequences of spatial heterogeneity; heterogeneity is a Refers to the tangential force per unit area exerted by a fluid neutral theory is a first approximation to measure of how different parts of a landscape are from as it moves across a surface. Shear stress causes erosion and ecological reality, or a basic theory that each another. This discipline also examines how spatial sloughing of biofilms, and initiates the formation of the provides the essential ingredients to further structure affects organisms. filamentous streamers characteristic of many biofilms. explore theories that involve more complex 33 Mass transfer Stochastic theory assumptions. Along the same lines, Chave The transport of dissolved molecules and particles from Stochastic theory deals with processes that are non- emphasizes that neutral and niche theories the bulk liquid into the biofilm. Diffusion is the process by deterministic in that the next state of the environment is are complementary, not conflicting, and which molecules move from areas of higher concentration partially but not fully determined by the previous state of that this link should be contained in future to areas of lower concentration. Advection is the process the environment; or, stochastic processes are largely by which moving liquid actively carries molecules and governed by chance alone. By contrast, a deterministic improvements of community ecology. In the particles. process is causally determined by an unbroken chain of following sections, we consider the interplay prior occurrences. between neutral and deterministic processes, Metacommunity and identify and describe dispersal, invasion A metacommunity is a set of local communities that are Tensoactive rhamnolipids linked by the dispersal of multiple, potentially interacting, Rhamnolipids are tensoactive glycolipids that contain one and the possible role of cell–cell signalling species. or two rhamnose molecules. in the invasion of biofilms as processes that

PERSPECTIVES

potentially influence community assembly valuable tool for the prediction of invasive spread better in fragmented landscapes in and that add further complexity to our spread, and for understanding the which patches can create ‘stepping stones’ for neutral ‘null hypothesis’. These processes mechanisms underlying spread dynamics18. dispersal40. essentially determine the immigration rate, Mechanisms underlying dispersal patterns Based on these concepts, we postulate which, in neutral models, can function as a are numerous and complex. For instance, that the interplay between biofilm topogra- ‘barrier’ or ‘filter’ for community assembly31. structural features of the forest canopy can phy, fluid dynamics and propagule proper- affect fluid dynamics above the canopy, ties shape spread dynamics in microbial Dispersal in microbial landscapes which in turn affects seed dispersal patterns18. landscapes. Two different scenarios with Recent advances in time-lapse microscopy Furthermore, poor dispersers are predicted different dispersal kernels show this inter- have revealed the various modes of dispersal to spread better in landscapes in which play (FIG. 1d). In heterogeneous microbial of biofilm organisms, such as rippling, roll- fragmentation is low, whereas good dispersers landscapes in which topography is ing or detachment3. These dispersal modes mainly result from physical forces and are therefore passive. By contrast, evidence from Box 1 | Neutral community models monospecies biofilms shows that bacteria can a c

) 4 ) also leave biofilms by several mechanisms, N 4 A A B A A B N including cellular differentiation, which 2 2 results in cells that have different functions B C C 0 B A C Number of species, S( Number of 0 and fates, matrix dissolution and the induc- 1234 species, S( 34–37 Abundance, N 1234 tion of motility functions . For instance, Abundance, N prophage-mediated apoptosis-like death of An individual dies ) 4 biofilm cells was proposed to constitute an A A B N b important mechanism of differentiation that 2 B C C facilitates the liberation of a subpopulation of A A B Number of 0 species, S( cells by local lysis and the formation of motile 1234 A new assembly Abundance, N 33,37 B C dispersal cells (FIG. 1c). Alterations in is created almost substrate availability can also induce similar immediately. ) 4 A A B N dispersal behaviour, and both transcriptom- 2 ics and proteomics have shown that the B B C

Number of 0

differentiation of specialized dispersal cells is species, S( 34,35 1234 highly regulated . Proteomics also showed Abundance, N that dispersing cells of Pseudomonas aerugi-

) 4 nosa are more similar to planktonic cells than A A B N 34 to mature biofilm cells , which indicates that 2 dispersing biofilm cells revert to the plank- B D C

Number of 0 species, S( tonic mode of growth. The confirmation of 1234 detachment as an intrinsic behaviour in bio- Abundance, N films has led to the appreciation of dispersal as an insurance policy to seed a new biofilm In common with the theory of island biogeography60, neutral community models recognize a during resource limitation (or simply during ‘mainland’ community that functions as a source pool for immigrants. Immigrants are taken randomly from this pool with their relative abundance being the sole arbiter of the probability that ageing) of the parental biofilm6,36,37. they become established in the local community. The degree of isolation of the local community Although detailed mechanisms that dictates the probability that an immigrant, rather than a locally reproduced organism, replaces a regulate dispersal have been identified, dead individual in the local community. By attributing all the dynamics to chance and ignoring microbiologists lack conceptual models to factors such as competition for resources, niche differentiation, disturbance, food webs and a predict the spread dynamics of propagules. multitude of other mechanisms that most ecologists consider should be represented in any Understanding spread dynamics is crucial for comprehensive theory, Hubbell24 and Bell25 undoubtedly courted controversy when they proposed a dispersal-assembled theory such as the neu- a neutral community model to explain biological community structure. However, despite its tral theory26, but also applies to understanding simplicity, the model can simulate community structures, in the form of taxa-abundance pathogen dispersal and disease gradients38. distributions, that are remarkably similar to those observed in plant communities in tropical forests, Microbiologists need experimentally testable as well as communities of birds, insects, fish and bacteria. The figure shows a schematic of neutral dynamics in a small community, with only 6 individuals, models based on conditions and cellular initially comprising 3 equally abundant taxa (part a). Hubbell assumes that communities are traits for habitat selection, as derived from saturated with individuals. This means that for an assemblage to change, an individual must die or landscape ecology and dispersal ecology. One leave the system (part b). It is then immediately replaced by an individual, either by reproduction in prediction from these models is that dispersal the community or by an immigrant from a source community (part c). Therefore the community not only determines the potential area of forms and develops through a continuous cycle of immigration, reproduction and death. propagule recruitment, but also functions as It is not only the relative abundance of taxa at a single site that is influenced by stochasticity and a template for post-dispersion processes, such immigration, but also spatial patterns in relative abundance. Both Hubbell and Bell demonstrated as gene flow, competition, invasion and the the emergence of clusters in the spatial distribution of taxa by applying their neutral model to an assembly of new communities39. The spatial array of local communities on a regular grid and allowing random dispersal between neighboring distribution of propagules can be represented grid cells. The simulated decay in the similarity of community structure with the distance between microbial cells is similar to that observed in intensive surveys of trees, shrubs, herbs and birds. by the dispersal kernel model, which is a

PERSPECTIVES

characterized by protuberances and voids, resident cells and propagules. Coaggregation hydrated. Diffusion is pathway-controlled16, flow fields could affect the fate of prop- occurs when several bacterial species bind and chemical and physical heterogeneity agules. By analogy with the fluid dynamics to each other through the interaction of can therefore affect signal transmission. For over vegetation canopies18, propagules cell-surface-expressed molecules, such as instance, the chemistry of extracellular poly- might leave the main current and, assisted lectin–saccharide coupling, and is thought meric substances is variable over small scales54 by swimming, deposit through the bot- to produce the complex sequential structure and produces environmental gradients that tom-ward trajectories of micro-eddies. of dental and aquatic bacterial biofilms46. might function as boundaries, filtering or Streamers that are particularly exposed The impact of the spatial organization deflecting signals. By contrast, the advective to the flow might function as hotspots for that results from these interactions on the flow of water through channels might dilute propagule landing, as has been shown for invasion of biofilms and community assem- the pool of signalling molecules but could suspended particles in stream biofilms11. bly has not been investigated. We propose also constitute a fast transmission lane. This would result in ‘short-distance’ or that the response of propagules to a new Based on these considerations of signal ‘intermediate’ dispersal, and in a dispersal environment in the microbial landscape propagation, we suggest that signalling by kernel with a typically reduced recruitment depends on the mode of dispersal and phe- resident cells in response to invasion as some area. Flow fields over a less heterogeneous notypic variation. Dispersed cell clusters that sort of defence mechanism would have to landscape would induce ‘long-distance’ are enclosed in matrix fragments might not occur close to the preferential invasion site in dispersal and larger recruitment areas. travel far, but after contact with ‘barren land’ order to be effective. This theory is supported Collectively, different dispersal capa- or resident biofilms, they are better-suited by the quorum-sensing-induced production bilities and landscape patterns might to establish a new biofilm than single cells. of tensoactive rhamnolipids in channels, which differentially affect dispersal-assembled Increased numbers of cells and the residual were at first thought to prevent clogging communities. For instance, poor dispersal matrix might transiently buffer propagule of channels in P. aeruginosa biofilms55, but capabilities of rare species in the source com- cells against hostile environmental conditions. might also be a strategy to reduce the inva- munity (BOX 1) mean that these species will Single motile cells can actively seek suitable sion of these biofilms56. Channels and pores be under-represented in local communities environments but are more susceptible to in the biofilm matrix might be preferential and might become extinct. Therefore, rare grazing by protozoa than clusters of cells. invasion routes into biofilms, as propagules species might take longer to re-establish than might be transported through them by advec- more abundant species41. This differential Cell-to-cell signalling. The relationship of tion, in a similar fashion to the function of barrier to dispersal is evidently important cell–cell signalling47–49 to the colonization of rivers as conduits in terrestrial landscapes. for the formation of diverse local communi- microbial landscapes might also be relevant, Propagules might also preferentially land on ties31, and for the spatial heterogeneity of as bacterial communication might enable canopy structures such as mushroom caps diversity in microbial landscapes. self-recognition, intentional behaviour, or streamers that are exposed to the bulk ‘decision-making’ at the group-level and liquid flow. In this case, signalling by residents Invasibility in microbial landscapes the recognition and identification of other close to the surface of these structures would Once a propagule has successfully dispersed colonies50. For instance, cell–cell signalling optimize signal transmission to invading it can colonize barren land or invade a resi- might allow landing propagules to dif- cells. However, signal molecules would also dent community. Factors that influence the ferentiate between ‘self’ and ‘non-self’ (for be highly susceptible to flow-induced loss. invasibility (that is, the ease of invasion) of a example, resident cells), and bacteria might This scenario complements the conceptual resident community include the fitness of the even have the ability to determine the ratio model that was proposed by Parsek and resident organisms, their diversity, the availa- of self-to-non-self and modulate phenotypes Greenberg57, which predicts that the highest ble resources and the spatial configuration of accordingly48. Of equal importance in a new signal concentration occurs where biomass these factors in the landscape42,43. Therefore, environment is the notion that signalling accumulation best protects signal molecules at this stage of dispersal-driven community molecules provide information on the diffu- from flow-induced loss (for example, at the assembly, deterministic processes related sion and mixing environment around cells centre of mushroom caps). It is possible that to interactions between newcomers and and might function to control the expression cells expressing a ‘guardian’ phenotype that is residents, and to coexistence through niche or repression of extracellular enzymes51. specialized for cost-intensive signal produc- differentiation, might become important How signalling molecules propagate tion colonize the peripheral parts of biofilms. for the formation of biofilms and the land- through microbial landscapes remains poorly This would constitute an altruistic strategy scapes that they form. Interactions between understood, yet it is of prime importance that benefits the group by reducing the newcomers and residents can potentially if the suppression of signalling is to be invasion of the biofilm. range from commensalism and mutualism to used to eradicate biofilms from indwelling More generally, incoming propagules parasitism, and these interactions contribute devices or remove contaminating biofilms. might face chemical barriers synthesized to the non-random spatial distribution Signalling molecules seem to function over by resident bacteria, such as antibiotics and that is observed in microbial landscapes44. short distances (on the order of 1 µm) in other toxins. These molecules seem to have For instance, Maixner and colleagues45 open systems such as dental plaque biofilms an essential role in determining microbial showed that resource gradients in nitrifying that are grown in flow cells52, but long-range diversity58 and in the structure of biofilm biofilms can induce niche differentiation of ‘calling-distances’ (mostly 4–5 µm, rang- communities. Competition experiments with coexisting Nitrospira populations (FIG. 1e). ing up to 78 µm) have been identified in a marine epiphytic biofilm confirmed this, However, no one has addressed whether the rhizosphere environment53. Signalling as strains that produce antibacterial extra- this coexistence might affect the invasion of molecules are typically small and, depending cellular compounds out-compete or invade biofilms. Furthermore, coaggregation is an on their hydrophobicity, could diffuse freely biofilms of strains that do not produce such important example of an interaction between through the biofilm matrix, which is highly inhibitors59.

PERSPECTIVES

Perspectives 12. Leibold, M. A., et al. The metacommunity concept: a 42. Shea, K. & Chesson, P. Community ecology theory as a framework for multi-scale community ecology. Ecol. framework for biological invasions. Trends Ecol. Evol. Microbial ecologists need theories that can Lett. 7, 601–613 (2004). 17, 170–176 (2002). draw together simple models and concepts. 13. Konopka, A. Microbial ecology: Searching for 43. Loreau, M., Mouquet, N. & Gonzalez, A. Biodiversity principles. Microbe 1, 175–179 (2006). as spatial insurance in heterogeneous landscapes. We are optimistic that revisiting classical 14. Turner, M. Landscape ecology: What is the state of the Proc. Natl Acad. Sci. USA 100, 12765–12770 ecology can herald a new era in biofilm science? Annu. Rev. Ecol. Evol. Syst. 36, 319–344 (2003). (2005). 44. Tolker-Nielsen, T. & Molin, S. Spatial organization of research that will enable microbial ecol- 15. Reiners, W. A. & Driese, K. L. Transport Processes in microbial biofilm communities. Microb. Ecol. 40, ogy to further develop as a field, and will Nature (Cambridge University Press, Cambridge, 75–84 (2000). 2004). 45. Maixner, F. et al. Nitrite concentration influences the bring microorganisms closer to the trees in 16. Lovett, G. M., Jones, C. G., Turner, M. G. & Weathers, population structure of Nitrospira-like bacteria. the forest. Our framework can contribute K. C. Ecosystem Function in Heterogeneous Environ. Microbiol. 8, 1487–1495 (2006). Landscapes. (Springer, New York, 2005). 46. Rickard, A. H., Gilbert, P., High, N. J., Kolenbrander, to a better understanding of how biofilm 17. Matz, C. et al. Biofilm formation and phenotypic P. E. & Handley, P. S. Bacterial coaggregation: an communities are established. It is our hope variation enhance predation-driven persistence of integral process in the development of multi-species Vibrio cholerae. Proc. Natl Acad. Sci. USA 102, biofilms. Trends Microbiol. 11, 94–100 (2003). that once the pathways to biofilm develop- 16819–16824 (2005). 47. Fuqua, C. & Greenberg, E. P. Listening in on bacteria: ment are understood, it will be possible to 18. Nathan, R. & Katul, G. G. Foliage shedding in acyl-homoserine lactone signaling. Nature Rev. Mol. deciduous forests lifts up long-distance seed dispersal Cell Biol. 3, 685–695 (2002). manipulate the growth of biofilms. Biofilms by wind. Proc. Natl Acad. Sci. USA 102, 8251–8256 48. Waters, C. M. & Bassler, B. L. Quorum sensing: cell-to- might eventually function as model systems (2005). cell communication in bacteria. Annu. Rev. Cell. Dev. 19. Rice, S. K., Collins, D. & Anderson, A. M. Functional Biol. 21, 319–346 (2005). to address emerging questions surrounding significance of variation in bryophyte canopy structure. 49. Keller, L. & Surette, M. G. Communication in bacteria: invasion and landscape ecology that are Am. J. Bot. 88, 1568–1576 (2001). an ecological and evolutionary perspective. Nature 20. Belnap, J., Hawkes, C. V. & Firestone, M. K. Rev. Microbiol. 4, 249–258 (2006). inaccessible through field observations of Boundaries in miniature: two examples from soil. 50. Ben Jacob, E., Becker, I., Shapira, Y. & Levine, H. communities of larger organisms. BioScience 53, 739–749 (2003). Bacterial linguistic communication and social 21. Picioreanu, C., Xavier, J. B. & van Loosdrecht, M. C. M. intelligence. Trends Microbiol. 12, 366–372 (2004). 51. Redfield, R. J. Is quorum sensing a side effect of Tom J. Battin is at the Department of Freshwater Advances in mathematical modeling of biofilm diffusion sensing? Trends Microbiol. 10, 365–370 structure. Biofilms 1, 337–349 (2004). Ecology, University of Vienna, Althanstrasse 14, (2002). 22. Hughes-Martinez, J. B. et al. Microbial biogeography: A-1090 Vienna, Austria, and Biofilm and Ecosystem 52. Egland, P. G., Palmer, R. J. & Kolenbrander, P. E. putting microorganisms on the map. Nature Rev. Interspecies communication in Streptococcus gordonii- Research Group, WasserCluster Lunz, Austria. Microbiol. 4, 102–112 (2006). Veillonella atypica biofilms: Signaling in flow 23. Dolan, J. R. Microbial biogeography? J. Biogeogr. 33, William T. Sloan is at the Department of conditions requires juxtaposition. Proc. Natl Acad. Sci. 199–200 (2006). USA 101, 16917–16922 (2004). Civil Engineering, Rankine Building, 24. Hubbell, S. P. The unified neutral theory of biodiversity 53. Gantner, S. et al. In situ quantification of the University of Glasgow, G12 8QQ UK. and biogeography. (Princeton University Press, spatial scale of calling distances and population 2001). density-independent N-acylhomoserine lactone- Staffan Kjelleberg is at the School of Biotechnology 25. Bell, G. Neutral macroecology. Science 293, mediated communication by rhizobacteria colonized and Biomolecular Sciences and Centre for Marine 2413–2418 (2001). on plant roots. FEMS Microbiol. Ecol. 56, 188–194 Biofouling and Bio-Innovation, The University of New 26. Alonso, D., Etienne, R. S. & McKane, A. J. The merits of (2006). neutral ecology. Trends Ecol. Evol. 21, 451–457 (2006). South Wales, Sydney, 2052 Australia. 54. Neu, T. R, Swerhone, G. D. W. & Lawrence, J. R. 27. Green, J. L., et al. Spatial scaling of microbial Assessment of lectin-binding analysis for in situ Holger Daims is at the Department of Microbial eukaryote diversity. Nature 432, 747–750 (2004). detection of glycoconjugates in biofilm systems. Ecology, University of Vienna, Althanstrasse 14, 28. Horner-Devine, M. C., Lage, M., Hughes, J. B. & Microbiology 147, 299–313 (2001). A-1090 Vienna, Austria. Bohannan, B. J. M. A taxa-area relationship for 55. Davey, M. E., Caiazza, N. C. & O’Toole, G. A. bacteria. Nature 432, 750–753 (2004). Rhamnolipid surfactant production affects biofilm Ian M. Head and Tom P. Curtis are at the School of Civil 29. Sloan, W. T. et al. Quantifying the roles of immigration architecture in Pseudomonas aeruginosa PAO1. Engineering and Geosciences, Cassie Building, and chance in shaping prokaryote community J. Bacteriol. 185, 1027–1036 (2003). University of Newcastle upon Tyne, NE1 7RU UK. structure. Environ. Microbiol. 8, 732–740 56. Espinosa-Urgel, M. Resident parking only: (2006). rhamnolipids maintain fluid channels in biofilms. Leo Eberl is at the Institute of Plant Biology, 30. Sloan, W. T., Woodcock, S., Lunn, M., Head, I. M. & J. Bacteriol. 185, 699–700 (2003) University of Zürich, CH-8008 Zürich, Switzerland. Curtis, T. P. Modeling taxa-abundance distributions in 57. Parsek, M. R. & Greenberg, E. P. Sociomicrobiology: microbial communities using environmental sequence the connections between quorum sensing and Correspondence to T.J.B. data. Microb. Ecol. (in the press). biofilms. Trends Microbiol. 13, 27–33 (2005). e-mail: [email protected] 31. Curtis, T. P. et al. What is the extent of prokaryotic 58. Lenski, R. E. & Riley, M. A. Chemical warfare from an diversity. Phil. Trans. R. Soc. B 361 2023 – 2037 ecological perspective. Proc. Natl Acad. Sci. USA 99, doi:10.1038/nrmicro1556 (2006). 556–558 (2002). 32. Houchmandzadeh, B. & Vallade, M. Clustering in neutral 1. Kolter, R. Surfacing views of biofilm biology. Trends 59. Burmolle, M. et al. Enhanced biofilm formation and ecology. Phys. Rev. E 68, 061912 (2003). Microbiol. 13, 1–2 (2005). increased resistance to antimicrobial agents and 2. Costerton, J. W., Lewandowski, Z., Caldwell, D. E., 33. Chave, J. Neutral theory and community ecology. Ecol. bacterial invasion are caused by synergistic Korber, D. R., Lappin-Scott, H. M. Microbial biofilms. Lett. 7, 241–253 (2004). interactions in multispecies biofilms. Appl. Environ. Ann. Rev. Microbiol. 49, 711–745 (1995). 34. Sauer, K., Camper, A. K., Ehrlich, G. D., Costerton, Microbiol. 72, 3916–3923 (2006). 3. Hall-Stoodley, L., Costerton, J. W. & Stoodley, P. J. W. & Davies, D. G. Pseudomonas aeruginosa 60. MacArthur, R. H. & Wilson, E. O. The Theory of Island Bacterial biofilms: from the natural environment to displays multiple phenotypes during development Biogeography. (Princeton University Press, 1967). infectious diseases. Nature Rev. Microbiol. 2, 95–108 as a biofilm. J. Bacteriol. 184, 1140–1154 (2004). (2002). Acknowledgements 4. Davey, M. E. & O’ Toole, G. A. Microbial biofilms: from 35. Sauer, K. et al. Characterization of nutrient-induced This article reflects the ideas of COMIX, a Collaborative ecology to molecular genetics. Microbiol. Mol. Biol. dispersion in Pseudomonas aeruginosa PAO1 biofilm. Research Project of the European Science Foundation Rev. 64, 847–867 (2000). J. Bacteriol. 186, 7312–7326 (2004). (EuroDIVERSITY); support came from a FWF Project. 5. Watnick, P. & Kolter, R. Biofilm, city of microbes. 36. Webb, J. S. et al. Cell death in Pseudomonas G. Pucher produced the microbial landscape figure; J. Webb, J. Bacteriol. 182, 2675–2676 (2000). aeruginosa biofilm development. J. Bacteriol. 185, F. Maixner, I. Hödl and B. Houchmandzadeh provided 6. Webb, J. S., Givskov, M. & Kjelleberg, S. Bacterial 4585–4592 (2003). material for the other figures. biofilms: prokaryotic adventures in multicellularity. 37. Webb, J. S., Lau, M. & Kjelleberg, S. Bacteriophage Curr. Opin. Microbiol. 6, 578–585 (2003). and phenotypic variation in Pseudomonas aeruginosa Competing interests statement 7. Kreft, J. U. Conflicts of interest in biofilms. Biofilms 4, biofilm development. J. Bacteriol. 186, 8066–8073 The authors declare no competing financial interests. 265–276 (2004). (2004). 8. Parsek, M. R. & Fuqua, C. Biofilms 2003: emerging 38. Hall-Stoodley, L. & Stoodley, P. Biofilm formation and themes and challenges in studies of surface-associated dispersal and the transmission of human pathogens. DATABASES microbial life. J. Bacteriol. 186, 4427–4440 (2004). Trends Microbiol. 13, 7–10 (2005). The following terms in this article are linked online to: 9. McFeters, G. A., Stuart, S. A. & Olson, S. B. Growth of Entrez Genome Project: http://www.ncbi.nlm.nih.gov/ heterotrophic bacteria and algal extracellular products 39. Nathan, R. & Muller-Landau, H. C. Spatial patterns of seed dispersal, their determinants and consequences entrez/query.fcgi?db=genomeprj in oligotrophic waters. Appl. Environ. Microbiol. 35, Pseudomonas aeruginosa 383–391(1978). for recruitment. Trends Ecol. Evol. 15, 278–285 10. Lock, M. A., Wallace, R. R., Costerton, J. W., Ventullo, (2000). FURTHER INFORMATION 40. With, K. A. The landscape ecology of invasive spread. R. M. & Charlton, S. E. River epilithon (biofilm): toward Tom J. Battin’s homepage: www.wasserkluster-lunz.ac.at a structural functional model. Oikos 42, 10–22 (1984). Conserv. Biol. 16, 1192–1203 (2002) . 11. Battin, T. J., Kaplan, L. A., Newbold, J. D. & Hansen, C. 41. Volkov, I., Banavar, J. R., He, F., Hubbell, S. P. & SUPPLEMENTARY INFORMATION Contributions of microbial biofilms to ecosystem Maritan, A. Density dependence explains tree species See online article: S1 (figure) processes in stream mesocosms. Nature 426, abundance and diversity in tropical forests. Nature Access to this links box is available online. 439–442 (2003). 438, 658–661 (2005).