Joseph E. Knelman

Patterns and Processes of Microbial Community Assembly

SUMMARY Recent research has expanded our understanding of microbial community assembly. However, the field of community ecology is inaccessible to many microbial ecologists because of inconsistent and often confusing terminology as well as unnecessarily polarizing debates. Thus, we review recent literature on microbial community assembly, using the framework of Vellend (Q. Rev. Biol. 85:183–206, 2010) in an effort to synthesize and unify these contributions. We begin by discussing patterns in microbial biogeography and then describe four basic processes (diversification, dispersal, selection, and drift) that contribute to community assembly. We also discuss different combinations of these processes and where and when they may be most important for shaping microbial communities. The spatial and temporal scales of microbial community assembly are also discussed in relation to assembly processes. Throughout this review paper, we highlight differences between microbes and macroorganisms and generate hypotheses describing how these differences may be important for community assembly. We end by discussing the implications of microbial assembly processes for ecosystem function and biodiversity.

Changes in assembly processes in soil bacterial communities following a wildfire disturbance

Although recent work has shown that both deterministic and stochastic processes are important in structuring microbial communities, the factors that affect the relative contributions of niche and neutral processes are poorly understood. The macrobiological literature indicates that ecological disturbances can influence assembly processes. Thus, we sampled bacterial communities at 4 and 16 weeks following a wildfire and used null deviation analysis to examine the role that time since disturbance has in community assembly. Fire dramatically altered bacterial community structure and diversity as well as soil chemistry for both time-points. Community structure shifted between 4 and 16 weeks for both burned and unburned communities. Community assembly in burned sites 4 weeks after fire was significantly more stochastic than in unburned sites. After 16 weeks, however, burned communities were significantly less stochastic than unburned communities. Thus, we propose a three-phase model featuring shifts in the relative importance of niche and neutral processes as a function of time since disturbance. Because neutral processes are characterized by a decoupling between environmental parameters and community structure, we hypothesize that a better understanding of community assembly may be important in determining where and when detailed studies of community composition are valuable for predicting ecosystem function.

Nutrient Addition Dramatically Accelerates Microbial Community Succession

The ecological mechanisms driving community succession are widely debated, particularly for microorganisms. While successional soil microbial communities are known to undergo predictable changes in structure concomitant with shifts in a variety of edaphic properties, the causal mechanisms underlying these patterns are poorly understood. Thus, to specifically isolate how nutrients – important drivers of plant succession – affect soil microbial succession, we established a full factorial nitrogen (N) and phosphorus (P) fertilization plot experiment in recently deglaciated (∼3 years since exposure), unvegetated soils of the Puca Glacier forefield in Southeastern Peru. We evaluated soil properties and examined bacterial community composition in plots before and one year after fertilization. Fertilized soils were then compared to samples from three reference successional transects representing advancing stages of soil development ranging from 5 years to 85 years since exposure. We found that a single application of +NP fertilizer caused the soil bacterial community structure of the three-year old soils to most resemble the 85-year old soils after one year. Despite differences in a variety of soil edaphic properties between fertilizer plots and late successional soils, bacterial community composition of +NP plots converged with late successional communities. Thus, our work suggests a mechanism for microbial succession whereby changes in resource availability drive shifts in community composition, supporting a role for nutrient colimitation in primary succession. These results suggest that nutrients alone, independent of other edaphic factors that change with succession, act as an important control over soil microbial community development, greatly accelerating the rate of succession.

Decreases in average bacterial community rRNA operon copy number during succession

Trait-based studies can help clarify the mechanisms driving patterns of microbial community assembly and coexistence. Here, we use a trait-based approach to explore the importance of rRNA operon copy number in microbial succession, building on prior evidence that organisms with higher copy numbers respond more rapidly to nutrient inputs. We set flasks of heterotrophic media into the environment and examined bacterial community assembly at seven time points. Communities were arrayed along a geographic gradient to introduce stochasticity via dispersal processes and were analyzed using 16 S rRNA gene pyrosequencing, and rRNA operon copy number was modeled using ancestral trait reconstruction. We found that taxonomic composition was similar between communities at the beginning of the experiment and then diverged through time; as well, phylogenetic clustering within communities decreased over time. The average rRNA operon copy number decreased over the experiment, and variance in rRNA operon copy number was lowest both early and late in succession. We then analyzed bacterial community data from other soil and sediment primary and secondary successional sequences from three markedly different ecosystem types. Our results demonstrate that decreases in average copy number are a consistent feature of communities across various drivers of ecological succession. Importantly, our work supports the scaling of the copy number trait over multiple levels of biological organization, ranging from cells to populations and communities, with implications for both microbial ecology and evolution.

Changes in community assembly may shift the relationship between biodiversity and ecosystem function

Can differences in community assemblyalter the relationship between biodiver-sity and ecosystem function? Pholchanet al. (2013) used a variety of manipu-lations to change microbial communityassembly in sludge reactors and exam-ined the subsequent links between diver-sity and a rare function, the removal ofendocrinedisruptingcompounds(EDCs).Interestingly, the authors saw no consis-tent differences between shifts in alphadiversity (e.g., species richness and even-ness) and ecosystem function, observingan increase, decrease and no difference inthe amount of removal of specific EDCswith increases in diversity. They suggestedthat differences in community assemblymay be driving variation in the rela-tionship between biodiversity and func-tion, a fascinating hypothesis that unitesprocesses in community and ecosystemecology.Combinations of four processes affectcommunity assembly: dispersal and diver-sification add new taxa to communitieswhile selection and drift affect their rela-tive abundances (Vellend, 2010; Nemergutet al., 2013). Particular research emphasishas been placed on assembly processesthataredriven bydifferences betweentaxa(“niche”) compared to those in whichany such differences are irrelevant to fit-ness(“neutral”)(Hubbell,2001).Likewise,researchers have focused on the roleof stochasticity, where assembly is moreprobabilistic vs. determinism, in whichrandomness does not affect communitydynamics.Nicheandneutralprocessescanoperateinunison(Adler et al., 2007)andboth can be affected by stochastic anddeterministic forces (Fox, 2012). Indeed,extensive data demonstrate that a varietyof factors, including nutrients, produc-tivity, resource availability, successionalstage, and disturbances may affect the rel-ative importance of different communityassembly mechanisms (Chase, 2007, 2010;Ferrenbergetal.,2013;Kardoletal.,2013).However, to our knowledge, no studieshave directly tested how shifts in com-munity assembly may affect the relation-ship between biodiversity and ecosystemfunction.Of course, a great deal of researchhas focused on pairwise combinationsof the interactions between communityassembly, biodiversity and/or function inisolation. First, a large body of workdemonstrates links between biodiversityand ecosystem function (Cardinale et al.,2011; Hooper et al., 2012), even formicrobial systems (Bell et al., 2005; Hsuand Buckley, 2009; Langenheder et al.,2010; Levine et al., 2011; Jousset et al.,2014). However, the nature and strengthof biodiversity ecosystem function (BEF)relationships have been widely debatedand strongly depend on the type of func-tion and ecosystem examined (Grime,1997; Hooper et al., 2005)andthedegree of redundancy within the com-munity (Reich et al., 2012; Jousset et al.,2014). These complexities may be height-ened for microorganisms due to theextraordinary phylogenetic diversity har-bored within microbial communities, andthe fact that a typical microbial commu-nity contains organisms from within avariety of functional guilds.Second, it is known that differentassemblymechanismsdrivebiodiversityindistinct ways. For example, spatial or tem-poral variation in environmental condi-tions increases biodiversity through nicheprocesses while increases in the diver-sity of the metacommunity or in theratio of immigration/emigration rates canincrease biodiversity through neutral pro-cesses (Vellend, 2010).Finally, a relatively new topic in the lit-erature relates community assembly andecosystem function (Fukami et al., 2010;Nemergut et al., 2013). Vital to such aconsideration is the relationship betweenresponse traits, or traits that can interactwithenvironmentalvariationtodeterminespecies distribution and abundance pat-terns, and effect traits, or traits that deter-mine the functional roles of different taxa(Naeem and Wright, 2003). When com-munities are largely structured by nicheprocesses, variation in the environmentcan directly correlate to effect traits thatare linked to selected response traits(Allison, 2012). However, when commu-nities are structured by neutral processes,ecosystem function will primarily dependon effect trait abundances within themetacommunity, dispersal and ecologicaldrift; thus, relationships between varia-tion in the environment and effect traits

Soil bacterial community structure remains stable over a 5-year chronosequence of insect-induced tree mortality

Extensive tree mortality from insect epidemics has raised concern over possible effects on soil biogeochemical processes. Yet despite the importance of microbes in nutrient cycling, how soil bacterial communities respond to insect-induced tree mortality is largely unknown. We examined soil bacterial community structure (via 16S rRNA gene pyrosequencing) and community assembly processes (via null deviation analysis) along a 5-year chronosequence (substituting space for time) of bark beetle-induced tree mortality in the southern Rocky Mountains, USA. We also measured microbial biomass and soil chemistry, and used in situ experiments to assess inorganic nitrogen mineralization rates. We found that bacterial community structure and assembly—which was strongly influenced by stochastic processes—were largely unaffected by tree mortality despite increased soil ammonium (NH4+) pools and reductions in soil nitrate (NO3−) pools and net nitrogen mineralization rates after tree mortality. Linear models suggested that microbial biomass and bacterial phylogenetic diversity are significantly correlated with nitrogen mineralization rates of this forested ecosystem. However, given the overall resistance of the bacterial community to disturbance from tree mortality, soil nitrogen processes likely remained relatively stable following tree mortality when considered at larger spatial and longer temporal scales—a supposition supported by the majority of available studies regarding biogeochemical effects of bark beetle infestations in this region. Our results suggest that soil bacterial community resistance to disturbance helps to explain the relatively weak effects of insect-induced tree mortality on soil N and C pools reported across the Rocky Mountains, USA.

Plant community and soil chemistry responses to long-term nitrogen inputs drive changes in alpine bacterial communities.

Bacterial community composition and diversity was studied in alpine tundra soils across a plant species and moisture gradient in 20 y-old experimental plots with four nutrient addition regimes (control, nitrogen (N), phosphorus (P) or both nutrients). Different bacterial communities inhabited different alpine meadows, reflecting differences in moisture, nutrients and plant species. Bacterial community alpha-diversity metrics were strongly correlated with plant richness and the production of forbs. After meadow type, N addition proved the strongest determinant of bacterial community structure. Structural Equation Modeling demonstrated that tundra bacterial community responses to N addition occur via changes in plant community composition and soil pH resulting from N inputs, thus disentangling the influence of direct (resource availability) vs. indirect (changes in plant community structure and soil pH) N effects that have remained unexplored in past work examining bacterial responses to long-term N inputs in these vulnerable environments. Across meadow types, the relative influence of these indirect N effects on bacterial community structure varied. In explicitly evaluating the relative importance of direct and indirect effects of long-term N addition on bacterial communities, this study provides new mechanistic understandings of the interaction between plant and microbial community responses to N inputs amidst environmental change.

Nutrient limitation of soil microbial activity during the earliest stages of ecosystem development

A dominant paradigm in ecology is that plants are limited by nitrogen (N) during primary succession. Whether generalizable patterns of nutrient limitation are also applicable to metabolically and phylogenetically diverse soil microbial communities, however, is not well understood. We investigated if measures of N and phosphorus (P) pools inform our understanding of the nutrient(s) most limiting to soil microbial community activities during primary succession. We evaluated soil biogeochemical properties and microbial processes using two complementary methodological approaches—a nutrient addition microcosm experiment and extracellular enzyme assays—to assess microbial nutrient limitation across three actively retreating glacial chronosequences. Microbial respiratory responses in the microcosm experiment provided evidence for N, P and N/P co-limitation at Easton Glacier, Washington, USA, Puca Glacier, Peru, and Mendenhall Glacier, Alaska, USA, respectively, and patterns of nutrient limitation generally reflected site-level differences in soil nutrient availability. The activities of three key extracellular enzymes known to vary with soil N and P availability developed in broadly similar ways among sites, increasing with succession and consistently correlating with changes in soil total N pools. Together, our findings demonstrate that during the earliest stages of soil development, microbial nutrient limitation and activity generally reflect soil nutrient supply, a result that is broadly consistent with biogeochemical theory.

Oligotrophic wetland sediments susceptible to shifts in microbiomes and mercury cycling with dissolved organic matter addition

Recent advances have allowed for greater investigation into microbial regulation of mercury toxicity in the environment. In wetlands in particular, dissolved organic matter (DOM) may influence methylmercury (MeHg) production both through chemical interactions and through substrate effects on microbiomes. We conducted microcosm experiments in two disparate wetland environments (oligotrophic unvegetated and high-C vegetated sediments) to examine the impacts of plant leachate and inorganic mercury loadings (20 mg/L HgCl2) on microbiomes and MeHg production in the St. Louis River Estuary. Our research reveals the greater relative capacity for mercury methylation in vegetated over unvegetated sediments. Further, our work shows how mercury cycling in oligotrophic unvegetated sediments may be susceptible to DOM inputs in the St. Louis River Estuary: unvegetated microcosms receiving leachate produced substantially more MeHg than unamended microcosms. We also demonstrate (1) changes in microbiome structure towards Clostridia, (2) metagenomic shifts toward fermentation, and (3) degradation of complex DOM; all of which coincide with elevated net MeHg production in unvegetated microcosms receiving leachate. Together, our work shows the influence of wetland vegetation in controlling MeHg production in the Great Lakes region and provides evidence that this may be due to both enhanced microbial activity as well as differences in microbiome composition.

Patterns of Soil Bacterial Richness and Composition Tied to Plant Richness, Soil Nitrogen, and Soil Acidity in Alpine Tundra

ABSTRACT Patterns of soil bacterial richness using operational taxonomic units (OTUs) and abundance of bacterial groups (phylum or class) were studied in relation to plant richness and soil characteristics in the alpine at Niwot Ridge, Colorado, U.S.A. The study used a landscape gradient and snow fence in addition to plots amended with nitrogen (N). Bacterial richness was not correlated with total soil carbon (C) or total soil N, but showed strong positive correlations with pH and corresponding correlations with metallic cation concentrations. Bacterial richness showed a strong negative correlation (r = -0.86) with soil acidity and declined 30% over the pH gradient of 6.0–4.5. Plant richness correlated with acidity (r = -0.70) and declined 50% over this gradient. Bacterial OTU richness was sensitive to acidity but not to N amendments. However, abundance of five bacterial groups responded positively to N, four responded negatively, and three groups exhibited no changes. In plots with additional snow, snow additions reduced OTU richness. However, when snow was included in an ANCOVA model with N and soil acidity, OTUs were not affected, suggesting that snow effects were largely captured by soil acidity changes. Bacterial richness was correlated with forb richness and cover, but causal relationships remain unresolved.