Taxonomic composition

The Bacteria and Archaea contain tremendous genetic diversity, and despite recent efforts, it is certain that the vast majority of variation in stream bacterial communities has yet to be quantified. Most attention has been given to the oxic, heterotrophic bacteria inhabiting leaf packs and surface sediments because of their important roles in many ecological processes (e.g., McNamara and Leff 20041, Olapade and Leff 2004). However, numerous cases of high concentrations of anaerobically produced metabolic products, such as methane (e.g., Jones et al. 1995) and N₂O (e.g., Beaulieu et al. 2008) have made it clear that low-O₂ habitats are reasonably common in streams. For example, bacterial anaerobic metabolism, with release of methane and use of non-O₂ electron acceptors, occurs deep within hyporheic sediments where O₂ can be depleted (Baker 2000, Morrice et al. 2000).

Our ability to determine taxonomic composition has grown dramatically (Leff and Lemke 1998), but techniques still offer very different levels of resolution. At the finest level, DNA sequencing and matching to known organisms provides the closest thing to identification of species, although the species concept transfers poorly to bacteria. At coarser levels, approaches, such as division-level probes, can be used to probe for target groups, e.g., α-, β-, and γ-proteobacteria (Kirchman et al. 2004, Logue et al. 2008). Last, several finger-printing approaches (denaturing gradient gel electrophoresis [DGGE], temperature gradient gel electrophoresis [TGGE]) can be used to assess the degree of similarity among a set of samples with no information on actual taxonomic composition (Nikolcheva et al. 2003, Das et al. 2007). For instance, terminal restriction-fragment length polymorphism (tRFLP) can be used to document seasonal recurrence of certain taxa and is particularly effective at describing shifts in composition when augmented with sequencing of some bands (Hullar et al. 2006).

An important step is to link specific taxa with the functions they carry out. The prerequisite is the existence of primers to enable identification of both the taxon and the gene of interest. For example, the abundance of certain NH₄⁺-oxidizing Archaea varies between 2 river systems (Herfort et al. 2009) showing, in this case, that the process might be maintained by redundant taxa.

By far, most information available on aquatic fungi is for the hyphomycetes that are responsible for leaf decomposition in wadeable streams. These fungi actually are the asexual stages of ascomycetes and basidiomycetes rather than a phylogenetically distinct group. These microorganisms produce readily visible and identifiable spores (conidia) that make feasible tracking their presence and abundance, as measured by reproductive output (e.g., Ingold 1942, Bärlocher and Kendrick 1974). Typically, their reproductive cycle begins with a period (d–wk) of hyphal growth to accumulate sufficient resources for spore production. Next, spore-containing structures are produced above the leaf surface, and spores are released to the water column. The ability to identify spores collected from water samples has made possible good surveys of hyphomycete presence. Clear community-level differences exist based on water chemistry, temperature, and other factors (Bärlocher 1992). The fruiting (sexual) stage of basidiomycetes might be significant for decay of the nonsubmerged parts of logs. Chytrid fungi are especially interesting because they can infect/kill amphibians, and thereby affect the distribution of an important top consumer (Whiles et al. 2006).

All major groups of protozoans (flagellates, ciliates, and amoeba) are abundant and can be significant grazers of other microbes in stream ecosystems (Ribblett et al. 2005). Protozoa occur at high densities (∼10⁶ individuals/g sediment) and are consumed by meiofauna (Bott and Borchardt 1999). Protozoa can be much more abundant in accumulations of OM and fine sediments than in other habitats (Gücker and Fisher 2003). Most protozoa do not have strong abilities to hold themselves in place during high flow, so they rely on sheltered habitats and small size to keep them within the low shear-stress boundary layer. Ciliate abundances can be as high as 10⁶ individuals/m² (Reiss and Schmid-Araya 2008), and the smaller flagellates probably reach much higher densities (Schmid et al. 2000).

Biogeography

Study of large-scale patterns in microbial communities has lagged behind similar studies of other taxonomic groups, partly because of methodological constraints but perhaps more importantly because of the idea that dispersal was so widespread that disjunct patterns were unlikely (Baas-Becking 1934). Many microbes have some form of resting stage that is protected from harsh environmental conditions (desiccation, temperature extremes) and, therefore, is capable of long-distance transport by winds, water, or in association with other organisms. However, medium- (kilometers to tens of kilometers) to large- (hundreds to thousands of kilometers) scale differences in microbial composition clearly exist (Reche et al. 2005, Crump et al. 2007, Findlay et al. 2008).

Most of these patterns probably are driven by variation in strong environmental selectors. For example, soil and stream pH affect bacterial communities (Fierer et al. 2007), and stream water chemistry has large effects on both bacterial and fungal colonizers of leaf litter (Harrop et al. 2009). Examples from the tundra of North America (Crump et al. 2007) and high-elevation lakes (Reche et al. 2005) indicate that dispersal limitation can cause bacterioplankton communities to differ among lakes. Perhaps the best summary of our present state of knowledge is that some strong selectors (pH, OM attributes) clearly exist, but these selectors are not universal drivers of real-world patterns in microbial composition. Dispersal limitation might be important, but when dispersal limitation is critical might not be a simple question of geographic distance.

At much finer scales, substrate quality is capable of altering relative growth rates of microbial taxa and should be manifested as differences in taxonomic composition among substrates (Findlay and Sinsabaugh 2006, Judd et al 2006). However, no consistent body of evidence indicates differentiation of microbial communities among leaf types (see Das et al. 2007), although Fazi et al. (2005) found that the quality of OM in stream microhabitats strongly influenced bacterial composition.

Missing linkages

What is still missing from our understanding of microbial composition and distribution is knowledge of the functional links connecting measures of which groups are present, their inherent metabolic capacity, actual expression of that capacity, and consumption or release of biomolecules in the environment. In principle, tools targeted at each of these steps are available (see table 2 in Zak et al. 2006). However, the huge array of microbes, microhabitats, and processes in stream ecosystems complicates the decision regarding which section of the complex network to tackle first. Nevertheless, stream ecologists have a large range of processes and disciplinary points of view, and enough information exists that they should be able to generate interesting and ecologically relevant hypotheses about connections between composition, activity, and ecosystem function. I suggest that we begin by proposing which processes are most probably limited by availability of the appropriate taxa based on the number of taxa known to have a certain capability. For instance, the ability to use NO₃⁻ as an electron acceptor is widespread, so rates of denitrification are unlikely to be controlled by absence of denitrifying bacteria. However, the ability to degrade certain complex organic molecules, such as lignin, is less common, and organisms with this ability might not always be present despite the presence of appropriate substrate.

Biomass

Stream microbes occur in or on all surfaces and voids in stream habitats, and some reasonably predictable shifts in abundance and composition occur among locations within these habitats. In low-order streams, biomass in sediments or on surfaces exceeds the planktonic microbial biomass, but the relative contribution can change with season and flow (Edwards et al. 1990). Surface benthic sediments, including OM accumulations, harbor the highest microbial biomass (100–500 mg microbial C/m²) in most streams (Findlay et al. 2002).

The relative abundance of bacteria vs fungi is quite predictable in stream habitats. For instance, the relative abundance of fungi is much more sensitive to particle size than is bacterial abundance (Findlay et al. 2002), so bacterial biomass predominates on fine particulate OM and (generally) in deeper sediments (Findlay et al. 2002). However, even small quantities of coarse particulate OM allow fungal biomass to exceed bacterial biomass in a stream. Fungi dominate pooled microbial biomass on leaves and wood (e.g., Hieber and Gessner 2002), whereas bacteria dominate sediments and fine particles. In hyporheic sediments, which usually lack large accumulations of coarse particulate OM, the fungal contribution is small relative to bacterial biomass (Ellis et al. 1998, Crenshaw et al. 2002).

Fungal biomass makes up the largest part of heterotrophic microbial biomass in the first days to weeks of leaf litter decay (Findlay and Arsuffi 1989, Weyers and Suberkropp 1996). Therefore, patches of leaf litter within a stream or streams with high and sustained litter inputs can be hot spots of fungal abundance. However, fungal mass fluctuates rapidly and might be less stable (reliable) in stream ecosystems where litter inputs are periodic and litter is retained only for short periods.

Biofilms occur on stone, wood, and plant surfaces, and these highly diverse communities are responsible for significant primary production and C metabolism (Battin et al. 2008). Bacteria inhabiting epilithic biofilms can reach abundances of 10⁷ cells/cm² of surface area, and their abundance is probably limited by the C supply or time since disturbance. Biofilms proceed through a reasonably predictable growth phase with colonization and growth of bacteria and algae (Sobczak 1996) leading to a thickening of the layers until basal cells become starved of O₂, C, or nutrients because of poor supply from stream water flowing past the biofilm. Biofilms are susceptible to physical disruption by high flow, scouring by sediment particles in transit, or disturbance by grazing insects. The rate of biofilm accumulation varies dramatically among streams (Hudson et al. 1992) and over time within a stream (Findlay et al. 1993). The presence or absence of a significant autotrophic component strongly influences overall biofilm growth (Haack et al. 1988) and can alter amounts and pathways of C transfer and degradation within the biofilm (Rier et al. 2007).

Biofilms rely on delivery of gases, OM, and inorganic nutrients from flowing stream water, and they often have physical structures, such as macropores and streamers (Battin et al. 2003), that facilitate flux from the overlying water into the core of the biofilm itself. Both algae and bacteria produce extracellular material that helps provide some structure and could help adsorb dissolved organic matter (DOM) from the overlying water (Freeman and Lock 1995, Findlay and Sobczak 1996, Fisher 2003,) and allow extracellular enzymes to yield energetically valuable monomers.

Production

Productivity is the generation of new OM over time. New OM production might be the result of conversion of inorganic C into organic C (primary production) or conversion of existing organic C into a new organism (secondary production). Stream microbes are significant primary and secondary producers. Unicellular algae and cyanobacteria account for much of the photosynthetic autotrophic production in streams (Larned 2010), but microbial chemosynthetic autotrophic production also can be present. Perhaps the best known example occurs during nitrification when certain bacteria oxidize NH₄⁺ in the presence of O₂ to derive energy for fixation of CO₂. In addition to new production, the net result is transformation of NH₄⁺ to NO₃⁻, and this process has been well-studied (e.g., Kemp and Dodds 2002). Nitrifiers have low growth efficiencies, so the actual C production from this process is unlikely to be high in most streams. Moreover, heterotrophic microorganisms might out-compete nitrifiers for NH₄⁺ when C supply is high (Bernhardt and Likens 2002). Other avenues of chemosynthetic primary production include oxidation of sulfide by colorless and photosynthetic bacteria, annamox, and oxidation of ferrous iron. These reactions undoubtedly occur in stream sediments or where anoxic groundwaters reenter oxic environments (Baker et al. 2000), but the contribution to whole-system primary production is probably small in most streams.

Microbial secondary production at the expense of other organic materials is almost always a major part of whole-system secondary production. For instance, fungal production on decomposing leaves can be as much as 10% of litter input (Methvin and Suberkropp 2003). Aside from its absolute quantitative importance, microbial secondary production often converts otherwise unavailable OM into a form that can be ingested and assimilated by macroconsumers. Perhaps the most significant of these conversions is the uptake of DOM from solution and generation of new bacterial biomass (e.g., Wiegner et al. 2005). DOM is almost always the largest flux of organic C through a stream segment (Mulholland 1997), and unless it is converted to particulate OM via secondary production, DOM is unavailable to larger consumers. In the hyporheic zone, the influx of DOM can easily make up most of the C available (Sobczak and Findlay 2002). Uptake constants for individual identifiable DOM constituents can be large enough that DOM removal and temporary retention as microbial biomass is a significant fraction of downstream transport (Kaplan and Newbold 2003). Transfer to higher trophic levels is related to efficiency of removal from the overlying water and the proportion of assimilated C respired.

A fruitful area for future research might be to determine whether various pathways of production of microbial biomass are equally important in supporting higher trophic levels, e.g.: 1) Might geochemical conditions necessary for some chemosynthetic production limit grazer access to microbial biomass? 2) Is fungal production a more stable C source than rapidly cycling bacterial biomass? When seeking answers to such questions, the heterogeneity of stream ecosystems might be a strength rather than an impediment.

Progression

The first experimental demonstration that microbes played an important part in stream food webs might (perhaps) be traced to a seminal paper by Kaushik and Hynes almost 40 y ago (1971; Fig. 1). They observed that N accumulation in decomposing leaf litter was much slower when antifungal antibiotics were added to experimental treatments, a result suggesting that fungal growth was responsible for a significant portion of increases in leaf food quality. Moreover, consumers ate only small quantities of leaf material that had been autoclaved or aged in antibiotics relative to consumption of untreated leaf discs. Their findings and other similar studies (e.g., Iversen 1973) led to the view that microbial growth on litter was an essential prerequisite to consumption by larger consumers and that the microbial biomass was essentially the only digestible portion of leaf litter. This view was captured by Ken Cummins in his “peanut butter and crackers” analogy (Cummins 1974; Fig. 1), in which he suggested the microbial biomass was akin to peanut butter and the leaf substrate itself was indigestible cracker. Although probably not his intent, this effective analogy lead to a widely held paradigm that only the microbes were available to consumers and the leaf substrate contributed nothing. Numerous studies have shown associations between microbial colonization and apparent food quality (e.g., Arsuffi and Suberkropp 1984 [Fig. 1], Lawson et al. 1984 [Fig. 1], Graça et al. 2001), but answers to the question regarding the importance of microbes vs leaf material became more complex once estimates of actual microbial biomass became feasible (Findlay et al. 1986; Fig. 1). The quantity of bacterial and fungal biomass as C in the detritus–microbe complex is typically a only a small percentage of total leaf mass and reaches a maximum within a few weeks (e.g., Methvin and Suberkropp 2003). Therefore, absolute mass of substrate is so much greater than the mass of microbial C that leaf material might make the largest contribution to consumer C demand, despite the fact that the assimilation efficiency for microbial C is much greater (perhaps 5–10× higher) than assimilation efficiency for the nonliving leaf substrate. Selective feeding by larger consumers occurs at relatively large scales (i.e., one leaf vs another) and little opportunity exists to avoid consumption of the nonliving substrate. Therefore, assimilated C generally will be derived from both sources. However, smaller consumers might have finer-scale selective ability, such that they can feed on microbial biomass itself or on enriched patches on an individual leaf. Hall and Meyer (1998; Fig. 1) found that many smaller taxa could derive a significant portion of their C demand from bacterial biomass.

Microbes clearly play a role in altering detritus food quality, and microbes can have several important effects in addition to their contribution to consumer C needs. For example, aquatic hyphomycete fungi have differing abilities to degrade polymers present in leaf litter, and different hyphomycetes cause differential softening (predigestion?) of the detritus substrate (Arsuffi and Suberkropp 1984). Experiments in which leaf material is exposed to fungal enzymes show anticipated improvements in food quality (Bärlocher 1982; Fig. 1), a result suggesting the enzymatic effects of fungi on substrate digestibility might be as significant as their direct assimilation as microbial biomass.

In essence, our views of microbial roles in detrital food webs have changed dramatically in recent decades, although the basic observation that consumers prefer microbially conditioned leaf detritus has been confirmed numerous times. Detritus with high microbial biomass probably should be viewed as a cue for food quality rather than merely as a good source of microbial C. Stable-isotope approaches show that fairly low-biomass portions of the food web, such as benthic algae, might contribute disproportionately to higher trophic levels (Mayer and Likens 1987, McCutchan and Lewis 2002; Fig. 1).

The arguments presented above focus on C demands, which clearly are important for metabolic needs. However, N is frequently a limiting nutrient in consumer diets. N concentrations are much greater (perhaps 5×) in microbial biomass than in the substrate (the C concentrations are essentially the same in the 2 compartments). Thus, one could argue based on simple algebra that the proportional N contribution to consumers would be greater from microbial than from substrate biomass (Sanzone et al. 2001).

Inorganic nutrient availability can directly limit microbial growth, but recent perspectives on nutrient limitation have expanded to consider longer-term and whole-ecosystem consequences of nutrient enrichment (e.g., Chadwick and Huryn 2003; Fig. 1). Early researchers investigating the role of nutrients in leaf decay generally considered increased microbial activity as a positive effect (e.g., Suberkropp and Chauvet 1995). However, recent authors have asked how acceleration of microbial metabolism affects consumer secondary production over the longer-term. Results from a whole-stream fertilization experiment showed the expected stimulation of various microbial processes (Greenwood et al. 2007) and at least transient increases in invertebrate production (Cross et al. 2006; Fig. 1). However, clear signs also have been observed that the OM pools of the fertilized streams are being “mined” (Benstead et al. 2009; Fig. 1). Eventually, the decline in OM pools caused by faster microbial decomposition will cause decreases in heterotrophic growth, which will be followed by decreases in other processes dependent on detrital energy, such as denitrification or nutrient immobilization.

Thus, our view of microbes in detrital food webs has come full circle from an initial focus on their degradative activities (because those activities were easiest to measure) through a phase of directly tracking their transfer to consumers and now back to the larger consequences of how variation in rates and pathways of decomposition affect the whole stream ecosystem (see Gessner et al. 1999; Fig. 1).

Making microbes relevant to ecology

A significant divide exists between ecology and microbial ecology despite the huge contribution of microbial processes to important element cycles and transformations. With some notable exceptions (Long and Morin 2005, Bell et al. 2005), a broad and fruitful interchange has not occurred about how knowledge of microbes contributes to general ecological knowledge or how microbial systems might be good test cases for answering questions. In some respects, this lack of exchange is understandable. For instance, the species concept applies poorly to many microbial groups, and questions about competition become irrelevant if the players can rapidly change their genotypes and phenotypic capabilities. However, rapid response times and the possibility for multiple response variables (growth, element transformation) should make many microbes powerful cases for testing ecological principles. Moreover, looming practical issues of stream management and response to climate change would benefit from consideration of microbial processes. At the same time, these future changes in the stream physicochemical template will provide another axis of variability. I have argued throughout this overview that embracing variability in controlling factors and processes will lead to a stronger integration of microbial ecology into stream ecology.

Consideration of most stream microbial processes is not presently extensive when scientists and managers choose options for stream restoration or modification. In part, this failure to consider microbial process occurs because the practices necessary to ensure positive outcomes for larger target organisms (fishes) or processes (bank stability) are so complex and expensive that justifying consideration of additional factors is difficult. Nevertheless, it seems reasonable to assess how current practices affect the large array of microbial processes necessary for stream function. For instance, physical structures intended to add diversity to flow conditions (boulders, logs, etc.) also retain OM and, thus, provide substrate for microbial populations (Entrekin et al. 2008). Such effects could be benefits of some particular action, and if anticipated, might increase justification for necessary expenses. Moreover, some stream improvement plans could have unanticipated negative consequences on microbial process critical to restoration success. Until we understand all of the connections and trade-offs of a restoration strategy, including those related to microbes, we cannot have a useful debate about the true ecological, economic, and social costs/benefits of our management activities.

In contrast to the relatively small numbers of studies that made direct examinations of microbes as model systems, some microbial processes have been used extensively as response variables in experimental manipulations of resource diversity. For instance, many studies have addressed how mixtures of leaf species decompose relative to decomposition of individual species. Patterns often are not predictable, and both increases and decreases in decomposition rates have been reported for mixtures relative to single species of leaves (Swan and Palmer 2004, Kominoski et al. 2007b, Lecerf et al. 2007). Thus, we are not presently prepared to predict the effect on mass loss, food quality, or nutrient immobilization when riparian zone vegetation and, therefore, leaf inputs to streams, becomes more or less heterogeneous.

Future temperature changes will have predictable effects on rates of most microbial processes (Rowe et al. 1996, Acuña et al. 2008), but altered atmospheric composition might have the greatest and most widespread consequences. Many experiments have shown dramatic effects of elevated CO₂ levels on leaf chemical composition and, therefore, litter C content and composition (Kominoski et al. 2007a, Rier et al. 2002). In general, higher CO₂ is expected to cause higher C:N ratios, which will lead to slower decay and a potential for greater immobilization of inorganic nutrients. Litter inputs are often significant sources of C to streams, but higher atmospheric CO₂ is likely to increase the magnitude and persistence of this terrestrial–aquatic subsidy.

Stream hydrology will be significantly altered if predicted changes in amount of precipitation or its frequency lead to changes in persistence of flow or high-flow events. Inundation pattern can alter microbial communities (Rees et al. 2006) and in situ decomposition (Langhans et al. 2008). Consequences of climate change will be far-reaching and probably complex, but our basic understanding of streams and factors controlling their ecological functions is solid enough to make reasonable predictions about change. Moreover, the widespread availability of important techniques (hydrodynamic models and powerful tracer approaches) should enable quantification of change over the next 25 y.

I would argue that we are no longer limited by methods (although significant challenges remain), and that inclusion of microbes in discussions of stream biodiversity, resiliency, and functioning is both feasible and necessary. Microbes play a central role in many stream processes, and researchers working on emerging issues in stream ecology will need to include microbial processes explicitly in conceptual and quantitative models of stream functioning.