OUP user menu

Insights into the resistance and resilience of the soil microbial community

Bryan S. Griffiths, Laurent Philippot
DOI: http://dx.doi.org/10.1111/j.1574-6976.2012.00343.x 112-129 First published online: 1 March 2013


Soil is increasingly under environmental pressures that alter its capacity to fulfil essential ecosystem services. To maintain these crucial soil functions, it is important to know how soil microorganisms respond to disturbance or environmental change. Here, we summarize the recent progress in understanding the resistance and resilience (stability) of soil microbial communities and discuss the underlying mechanisms of soil biological stability together with the factors affecting it. Biological stability is not solely owing to the structure or diversity of the microbial community but is linked to a range of other vegetation and soil properties including aggregation and substrate quality. We suggest that resistance and resilience are governed by soil physico-chemical structure through its effect on microbial community composition and physiology, but that there is no general response to disturbance because stability is particular to the disturbance and soil history. Soil stability results from a combination of biotic and abiotic soil characteristics and so could provide a quantitative measure of soil health that can be translated into practice.

  • biodiversity
  • decomposition
  • disturbance
  • ecosystem services
  • nitrogen cycling
  • soil health
  • soil structure
  • stability


Microorganisms living in soil are abundant and highly diverse, with estimates of up to 109 cells (Gans et al., 2005) and 104 species (Curtis et al., 2002) per gram of soil. These microorganisms are the key players of many soil functions such as biogeochemical cycling, plant productivity or climate regulation and are essential for the integrity of terrestrial ecosystems. Soil, which is a nonrenewable resource, is increasingly under environmental pressure most often related to the intensification of human activities (Creamer et al., 2010). Given the crucial importance of maintaining soil functions, there has been considerable effort invested in understanding the response of soil ecosystems to disturbance or environmental change and the resistance and resilience of soil microorganisms (Fig. 1). This increasing number of studies of soil microorganisms in relation to the functional stability of soil ecosystems is probably also due to the upsurge in research on the connections between biodiversity and ecosystem functioning (Loreau, 2010). Loss of biodiversity has been identified as a major threat to soil (COM(2006)231) but because of the complexity and variability of soil, microbial ecologists face a huge challenge in quantifying the role of microorganisms in enabling the soil to cope with disturbance and the underlying mechanisms remain poorly understood.

Figure 1

The response of the soil ecosystem to disturbance is influenced by the resistance and resilience of the soil microbial community.

Resistance and resilience are also considered as ecological concepts of high policy relevance, that is, ‘how can we increase the resilience of habitats and species to cope with climate change?' (Sutherland et al., 2006). Almost any study could be considered in terms of resilience, in that the common experimental format – the effect of X on Y – will have information about the effect of a disturbance on the system if X is a disturbance. To get an understanding of resilience, there needs to be a measurement soon after the disturbance to gauge resistance and then several subsequent measurements to assess the pattern of resilience. The time period can be a matter of days in a laboratory incubation, or even minutes for some physical measurements (Zhang et al., 2005), through to years for field-based observations and is generally related to the nature of the disturbance. Thus, there have been studies spanning hundreds of years, for a postglacial chronosequence (Orwin et al., 2006); tens of years for field-based restoration projects (Zhang et al., 2010); or generally from 3 to 60 days in microcosm experiments. Studies have investigated the resilience of microbial communities to disturbance due to human activities such as land use and agricultural practices but also to natural disturbance such as fire or freeze–thaw.

There is now a sufficient body of literature on the biological resilience of soils to draw a synthesis of the controlling factors and to examine the underlying mechanisms. This body of literature is summarized in Table 1 which groups studies according to the primary disturbance or the research theme. While resilience can refer to either a population or a function (Botton et al., 2006), the majority of studies reviewed in this study have included a microbial function (e.g. nitrification, denitrification and decomposition), while some have also simultaneously measured changes in microbial community structure or diversity.

View this table:
Table 1

Summary of soil microbial stability studies grouped by disturbance or research theme

EcosystemLocationSoil typeSoil treatmentSituationExperimental disturbanceDurationResistance and resilience ofMicrobial community measuresReference
FBRL (F)Forest or agricultureFHeat30 daysEnzyme activity, respirationPLFAChaer et al. (2009)
ANZSiCLGrassland or tillageFHCl; NaCl; Cu; D-W; F-T14 daysCmic, catabolic potential Degens et al. (2001)
AUSSLGrassland or tillageFD-W5 daysCmic, C & N dynamicsPLFASteenwerth et al. (2005)
A, PUKC, CL, SCL, SC, SL, LSGrassland or tillageFHeat; Cu; compression28 daysDecomposition, physical structureELFAGregory et al. (2009)
FUSORG over SiLBurningFWarming (in field plots)3 yearsCmic, enzyme activityClone & sequenceAllison et al. (2010)
FAUSBurning, topologyFHeat37 daysCmic, decompositionDGGE, PLFABanning & Murphy (2008)
PESDC & DRBurning, retardantFn/a1 yearBacterial & fungal growth, CLPPVelasco et al. (2009)
ATNSC (CLC)n/aLHeat +/− glucose56 daysRespirationHamdi et al. (2011)
Land management
AJPA, E, UOrg/Con fertilizerFFumigation, grinding56 daysCmic, decomposition; nematodesFujino et al. (2008)
AUKSL, SCL, CCompactionFCompression7 daysPhysical structureELFAGregory et al. (2007)
AUKLSOrg/Con cultivationFHeat; Cu60 daysDecomposition
PUKSCLReseed, sludge, biocide, N + limeFHeat; Cu28 daysDecompositionDGGEKuan et al. (2006)
FCNU, CSoil restorationFHeat; Cu; grinding28 daysDecompositionTRFLPZhang et al. (2010)
AUKSubsoil (DFC)Soil restorationGHeat; Cu; compression28 daysDecomposition, physical structureELFAGriffiths et al. (2008a)
VFRSLSterile/fresh compost1 yearCmic, SIRARISA, PLFASaison et al. (2006)
Metal contamination
PCNCopper (1 year)FCu40 daysCmic, SIR, PNR, qPCRPLFADeng et al. (2009)
ADKSLCopper (5 years)FCu, organic matter84 daysPICT, CLPP, respirationTRFLPBrandt et al. (2010)
VESC, RCopper (long-term)FCu (53 soils)1 dayPICT, growth rateFernández-Calviño et al. (2011)
ANLSCopper (long-term)FHeat; D-W60 daysDecomposition, bacterial growth rateTobor-Kaplon et al. (2005)
ANLSCopper (long-term)FPb; NaCl60 daysDecomposition, bacterial & fungal growth rateTobor-Kaplon et al. (2006b)
SBES, SLLime or compostFCu160 daysNitrificationKostov & Van Cleemput (2001)
AFRSiCLStraw or compost (15 years)FCuBacterial growth on Cu mediaClone & sequenceLejon et al. (2010)
CBESLZinc (10 years)FPesticide, D-W, F-T21 daysPNRMertens et al. (2007)
FNLSZinc (long-term)FHeat; NaCl; Pb60 daysDecomposition, bacterial & fungal growth rateTobor-Kaplon et al. (2006a)
PBESLLZn; NH4Cl1 yearPNRDGGERuyters et al. (2010a)
PBESLLZn, organic matter1 yearPDR, qPCR nosZDGGERuyters et al. (2010a, b)
CDKMercury (14 years)FHeat180 daysDecomposition, CLPPDGGEMüller et al. (2002)
AFRSiCHeat (35 °C), Cu, atrazineLHg150 daysARISABressan et al. (2008)
AFR EC Heat (35 °C), Cu, atrazineLHg120 daysNO3 reducing activityRFLPPhilippot et al. (2008)
PUKCL (EC)Metal contaminated sludgeFHeat; Cu; compression28 daysDecomposition, physical structureGriffiths et al. (2005)
SNOSLLCu, CD, Zn mixture60 daysPDR, toleranceHoltan-Hartwig et al. (2002)
CDESubsoil (SL)Petroleum contaminationFHeat; Cu60 daysDecompositionGriffiths et al. (2001a)
ADKSTylosinLHeat60 daysDecomposition, CLPPDGGEMüller et al. (2002)
ADKSLTylosin60 daysBacteria, fungi, protozoa, CLPPDGGEWestergaard et al. (2001)
SUSSiLL2,4,5-T42 daysTolerance, culture phenotypeDNA re-associationAtlas et al. (1991)
Soil properties
A, PUKSLMineral or organo-mineralFCu; benzene63 daysDecomposition, CLPPDGGEGirvan et al. (2005)
A, FUKCL; SContrasting soilsLHeat; Cu28 daysDecompositionPLFAGriffiths et al. (2008b)
A, P, F, MUKSC, SL, L, CL, S, SCL26 soils across ScotlandFHeat; Cu; compression28 daysDecomposition, physical structureKuan et al. (2007)
FCAC, TILLFD-W; Cu; HCl2 daysCmic, SIR, tolerancePLFARoyer-Tardif et al. (2010)
Biodiversity related studies
PUKSCLFum & inoculationLMatric potential96 daysCmic, catabolic potential, decompositionPLFADegens (1998)
PUKCLDifferential fumLHeat; Cu63 daysDecompositionDGGE, PLFAGriffiths et al. (2000)
PFRCLPlant species richnessGHeat; Cu60 daysDecompositionGriffiths et al. (2001a)
AUKCLSterile soil inoculatedLHeat; Cu60 daysDecompositionGriffiths et al. (2001b)
PUKSCLFum & inoculationLHeat; Cu28 daysCmic, decompositionDGGEGriffiths et al. (2004)
PNZn/aPlant species richnessFD-W16 monthsCmic, respiration, decompositionOrwin & Wardle (2005)
PFRSSterile soil inoculatedLHeat28 daysRespiration, PDR, PNRDGGEWertz et al. (2007)
Experimental studies
AUKCLn/aLDrying, sieving56 daysCmic, catabolic potential, decompDegens (1998)
ChrNZ, USPMVegetation chronosequenceFD-W3 daysCmic, respiration, decomposition, mineral NOrwin et al. (2006)
Al, ANP, SISL, LS, SiL, HLF-T60 cyclesEnzyme activity, respiration, qPCR 16SStres et al. (2010)
PNZPhysical disturbanceFn/a120 days454 sequencingLekberg et al. (2011)
  • Only those papers specifically referring to resistance and resilience are included in this summary table.

  • Ecosystem: A, tillage agriculture; AL, alpine; Chr, primary vegetation chronosequence; F, forest; M, moor; P, pasture (grassland); V, vineyard. Soil type: C, clay; L, loam; ORG, organic; S, sand; Si, silt; TILL, glacial till; A, andisol; C, cambisol; CLC, calcaric-leptic cambisol; DC, dystric-cambisol; DR, dystric-regosol; DFC, dystric-fluvic cambisol; E, entisol; EC, eutric cambisol; F, fragiudult; H, histosol; R, regosol; U, ultisol. Soil treatment: fum, fumigation; org/con, organic/conventional. Situation: F, field; G, glasshouse; L, laboratory. Experimental disturbance: D-W, dry–wet cycle; F-T, freeze–thaw cycle. Resistance and resilience of: CLPP, community level physiological profile; Cmic, microbial biomass carbon; Decomp, decomposition; PDR, potential denitrification rate; PICT, pollution induced community tolerance; PNR, potential nitrification rate; SIR, substrate induced respiration. Microbial community measures: ARISA, automated ribosomal intergenic spacer analysis; DGGE, denaturing gradient gel electrophoresis; ELFA, ester linked fatty acid; PLFA, phospho lipid fatty acid.

Resistance, resilience and functional stability


Throughout this review, we will adopt the definitions proposed by Rykiel (1985), that ‘disturbance’ is a biotic or abiotic cause which results in the effect of either a ‘perturbation’, response of an ecological component or process, or a ‘stress’, physiological response of an individual or functional response of the system. Definitions of resistance, resilience and stability in biological systems have been reviewed previously (Botton et al., 2006; Brand & Jax, 2007). Resistance is commonly defined as the ability of a system to withstand a disturbance, while definitions of resilience fall into two categories, engineering or ecological resilience (Fig. 2). Engineering resilience is where the behaviour of the system is treated like an engineering material that will show displacement and recovery towards its predisturbance state or a new stable state. Resistance to disturbance and the speed of return (resilience) are the two components of ecosystem stability as described by Pimm (1984), McNaughton (1994) and Loreau et al. (2002).

Figure 2

Schematic representation of engineering (a) and ecological resilience (b). In the engineering definition, stability is defined by the immediate response to disturbance (resistance) and then recovery over time (resilience). In the ecological definition, the ball in the basin represents the ‘state’ of the system. Resilience is a measure of how much disturbance can the system (ball) absorb so that it still remains in the same basin, before it flips over into another stable state (different basin).

Ecological resilience considers how much disturbance is required to move the system from one stable state to another alternate stable state, using the ‘ball and cup’ model (Holling, 1973; Gunderson et al., 2002; Fig. 1). Both definitions of resilience suffer from the difficulty of applying the concept of ‘stable state’ to natural ecosystems because they change not only in response to disturbance but are also subjected to gradual natural changes. However, Potts et al. (2006) were able to identify two stable states of ecosystem carbon and water fluxes following a wetting event, Gao et al. (2011) noted a degradation threshold for soil services at about 20% vegetation cover, while in experimental systems increasing disturbance did lead to alternative stable states in the community assembly of protists (Jiang & Patel, 2008). In a comprehensive review by Scheffer et al. (2001), analysis of several case studies suggested the presence of alternative stable states in various ecosystems but the authors also highlighted the difficulty of proving their existence, which required combinations of different approaches and modelling. There is a link between ecological and engineering resilience in that the theory of ‘critical slowing down’ proposes that recovery rates from small disturbances (i.e. engineering resilience) get slower and slower as a system approaches the tipping point between one stable state and another (i.e. ecological resilience) (Van Nes & Scheffer, 2007). This connection between engineering and ecological resilience has also been demonstrated for a range of complex systems, including: financial markets; epileptic seizure; fish stocks; catastrophic desertification and past climatic changes (Scheffer et al., 2009).

The engineering resilience approach predominates in the studies of soil biology and also in the study of soil physical parameters (Munkholm & Schjønning, 2004) and soil quality (Seybold et al., 1999). This review will therefore use the engineering definition of resilience; largely because experimentally it is convenient to consider the response and recovery of a population or function to a perturbation.

Calculating resistance and resilience

Calculations of resistance and resilience following disturbance can be performed by comparing between samples taken pre- and post-disturbance. However, such an approach does not take into account any changes occurring in control soils over time, either seasonal changes in the field or changes over time in an incubation. The use of control and disturbed samples at the same time eliminates this uncertainty and the calculation of resilience and resistance are usually presented as a proportion or percentage of the variable measured in the disturbed treatment to the control (undisturbed) treatment at the same time. Variations in the calculations are detailed in Table 2. Equations have generally taken the form of: change relative to the control (Sousa, 1980; Kaufman, 1982; Griffiths et al., 2000; Chaer et al., 2009); accounting for absolute differences between soils (Orwin & Wardle, 2004); or integrated measures (O'Neill, 1976; Fujino et al., 2008; Zhang et al., 2010).

View this table:
Table 2

Calculation of resistance and resilience indices

Embedded ImageEmbedded ImageKaufman (1982)
Embedded ImageEmbedded ImageSousa (1980)
Embedded ImageEmbedded ImageGriffiths et al. (2000)
Embedded ImageNCChaer et al. (2009)
Embedded Image Embedded Image Orwin & Wardle (2004)
Embedded Image Embedded Image Zhang et al. (2010)
NC Embedded Image O'Neill (1976)
  • C, variable measured in the control soil (undisturbed) at time 0 (immediately after disturbance) or at time x after disturbance; D, variable measured in disturbed soil at time 0 (immediately after disturbance) or at time x after disturbance; NC, not calculated.

Response of soil microorganisms to environmental disturbances

Changes occurring over primary successional gradients provided a means of testing several hypotheses regarding the interaction between soil nutrients and microbial stability (Orwin et al., 2006). These gradients covered the development of soils spanning hundreds of years but interactions, for example, between soil C, N, P status and the stability of microbial respiration to drying, depended on the location and the response variables considered. There were few consistent patterns of resistance and resilience and it was concluded that several factors might be affecting the outcome, such as: soil microbial community composition, plant community composition, adaptation to disturbance, soil texture and clay content, substrate diversity and food web structure.

Commonly studied environmental disturbances include, for example, fire, desiccation and freeze–thaw cycles. Adaptation to repeated freezing was demonstrated by comparing the response of Himalayan and temperate soils to repeated freeze–thaw cycles (Stres et al., 2010). Microbial respiration in temperate soils was far more susceptible to freezing than the Himalayan soils, but surviving temperate soil microorganisms eventually adapted to the changed environmental conditions. Adaptation was also evident in the stability of bacterial growth rates in Mediterranean pasture soils with high resistance and resilience to fire (Velasco et al., 2009). In a Mediterranean-type forest system subject to periodic burning, Banning & Murphy (2008) compared resilience of both a function (malate decomposition) and microbial biomass following a heat disturbance, in burnt or unburnt soil sampled from mounds or furrows. There were differences in both resistance and resilience related to nutrient limitation (soil C and N contents) as well as microbial biomass. They hypothesized that there was a threshold content below which the microbial biomass was no longer able to respond to the added substrate. Hamdi et al. (2011) also noted increased resistance of respiration to subsequent heat disturbance as a result of lower microbial biomass and reduced substrate availability. Similarly, Allison et al. (2010) suggested that nutrient limitation (depletion of labile C from the soil), resulting from burning of a boreal forest several years previously, increased the resistance of fungal community composition or microbial biomass to soil warming. This links with observations that C starvation increases microbial resistance to disturbance (Van Overbeck et al., 1995). In pure or mixed boreal forest stands resulting from wildfire, harvesting and soil type, resilience of the microbial biomass to experimental disturbances (dry–wet, Cu and HCl) was probably also related to nutrient limitation and thus to the parent geological material (Royer-Tardif et al., 2010). Resistance, on the other hand, was greatest in mixed stands, probably resulting from increased resource diversity leading to greater microbial diversity and more resistant taxa (Royer-Tardif et al., 2010). To study the effect of desiccation and rewetting, McKew et al. (2011) experimentally extended the desiccation period in a salt-marsh system and showed that while soil microbial function (extracellular enzyme activity) was resilient after flooding, microbial community structure was irreversibly altered with a different microbial community developing in the desiccated sediments compared with the control even after rewetting. The bacterial community composition was resistant to drying–rewetting cycles in grassland soils but not in an oak forest soil, while taxonomic diversity and richness were relatively insensitive to drying–rewetting frequency (Fierer et al., 2003).

Altogether, these findings partly reflect those of Orwin & Wardle (2005) that stability results from complex interactions but indicates that microbial community structure is not the sole determinant of functional resilience.

Response of soil microorganisms to management disturbances

As well as environmental disturbances experienced by many soil systems, disturbances often arise from land management. Tillage causes major changes to the soil system and direct effects on soil biology have been extensively documented (Wardle, 1995). For example, soil tillage resulted in significant modifications of the mycorrhizal fungal community structure (Jansa et al., 2002, 2003). Tillage also reduced the stability, compared with grassland soil, of C and N dynamics to a dry/wet event and caused a greater shift in microbial community structure (Steenwerth et al., 2005). Tilled soils also had functional characteristics that were less resistant to a range of disturbances (e.g. pH, osmotic, copper, dry/wet, freeze/thaw) than grassland soils, but this difference was independent of the size of the microbial biomass and more related to the functional characteristics of the community (the so-called catabolic response profile) (Degens et al., 2001). In a comparison of 15 soils from a range of long-term grassland and arable sites in the United Kingdom, grassland soils showed greater physical stability (to compression and wet/dry cycles) and biological stability (of plant decomposition to heat and copper disturbance) largely related to management effects on organic matter and interactions with clay content (Gregory et al., 2009). In tropical forest, sites converted to agriculture, enzyme activities were more stable to experimental heat disturbance in the forest soils, although there were clear differences in the stability depending on the functions monitored. Thus, a general function (i.e. fluorescein diacetate hydrolysis, which is a ubiquitous microbial activity) was equally stable in the forest and agriculture sites, while substrate-specific activities (i.e. cellulose and laccase which are not utilized by all microorganisms) were less resilient in the agriculture sites (Chaer et al., 2009). The conclusion was that the agriculture sites had a less diverse community of microorganisms capable of degrading the specific substrates than the unconverted forest sites. On the other hand, in temperate upland grassland plots where different management scenarios affected soil functional stability (of plant decomposition to heat and copper disturbance), there was no association between stability and broadscale changes in microbial community structure revealed by PCR-DGGE (Kuan et al., 2006). The authors suggested that there was a protective effect of certain soil properties and that microbial functions were more susceptible to a second disturbance following the initial disturbance represented by the agricultural practices (Kuan et al., 2006). Tobor-Kaplon et al. (2005, 2006a, b) tested this further in soils under long-term disturbance from copper, zinc or low pH. In soil collected from a pine forest, a transient heat disturbance imposed stronger changes in the stability of lucerne decomposition than lead or salt disturbances, but in soil of a similar texture from an arable site heat had the least effect on stability (Tobor-Kaplon et al., 2006a). However, in soil of a similar texture from an arable site under the influence of copper or low pH, a transient heat disturbance had the least effect on stability (Tobor-Kaplon et al., 2005, 2006b). This was related to the environmental fluctuations experienced by microorganisms at the two sites.

A practical application of heat disturbance is the use of soil solarization to decrease the incidence of disease in any subsequent crop. Although the vast majority of studies have concentrated on populations of specific pathogens, there has been some measurement of the effects on the broader soil microbial community structure. Culman et al. (2006) followed bacterial and fungal community structures by terminal restriction fragment length polymorphism (TRFLP) in a rice–wheat cropping system and reported clear and consistent effects of solarization (average temperature increase in 10 °C) only for fungi. Differences persisted into the next cropping cycle but there was resilience in that the communities became more similar over time. In a wheat–legume rotation with an average temperature increase in 20 °C, solarization was the main factor driving differences in the eubacterial community with similar but less pronounced effects for beta-proteobacteria, actinomycetes and alpha-proteobacteria (Gelsomino & Cacco, 2006). Bacterial diversity changed over time following solarization, increasing and then decreasing in this example, which might explain changes in functional resilience in soils given an experimental heat disturbance. These shifts in microbial community structure following solarization were not attributed to a direct thermal effect but rather to changes in the physico-chemical habitat or other ecological factors (Gelsomino & Cacco, 2006).

Organic amendments are often added to soils and can generally increase soil resilience. For example, Fujino et al. (2008) observed an increase in both resistance and resilience of cellulose activity to disinfection using sodium methyl dithiocarbamate (metam sodium) in soil amended with manure. Organically managed arable soils showed greater resilience of plant decomposition to experimentally applied heat or copper disturbances than intensively managed soils with no addition of organic matter (Griffiths et al., 2001a). Physical resilience of soil is also enhanced by the addition of organic matter, as demonstrated by the addition of peat to a clay soil (Zhang et al., 2005). Here, resistance to compression was actually reduced by the addition of peat but resilience increased and the authors related these fundamental observations to practical issues regarding the compaction of soils.

As a measure of the restoration of degraded soils, biological resilience took longer to improve than measures of the physical resilience of a subsoil (Griffiths et al., 2008a) which may indicate that a comparison of the two could indicate the stage of restoration reached.

Response of soil microorganisms to heavy-metal contaminated soils

At the extreme end of land management, soils may become contaminated with a variety of compounds such as heavy metals. A large body of microcosm and field studies provide evidence of a low resistance and resilience of most microbial functions to heavy metal. Copper is one of the most commonly investigated soil contaminants and so has attracted its fair share of stability studies. It has a broad-spectrum effect and alters the structure of the archaeal, bacterial and fungal communities (Wakelin et al., 2010a; Macdonald et al., 2011), with particular negative effects on Acidobacteria (Wakelin et al., 2010b; Macdonald et al., 2011) and positive effects on Actinobacteria (Lejon et al., 2008). Bacillus and Sphingomonas were particularly resistant to copper (Wakelin et al., 2010b). Brandt et al. (2010) concluded that the long-term response of microbial respiration to copper resulted from the ability of the microorganisms to develop Cu tolerance without affecting overall community structure. Lejon et al. (2010) suggested that copper adaptation was related to soil organic matter composition. The effects of an experimental, short-term, disturbance (Pb or NaCl) on the functional stability of soil from plots under a long-term disturbance from copper or low pH was dependent on the function measured (short-term decomposition of plant residues or the growth rates of bacteria and fungi) but was generally reduced by the long-term disturbance (Tobor-Kaplon et al., 2005).

The impact of heavy metals has been investigated on specific microbial guilds, in particular N-cycling microorganisms because nitrogen is the nutrient most often limiting for plant growth (Bollag & Barabasz, 1979; Bardgett et al., 1994). Kandeler et al. (1996) showed that enzymes involved in N-cycling were less resistant to increasing heavy-metal contamination than those involved in C-cycling. This is supported by the finding that decomposition of glucose was more stable than potential nitrification in a copper-contaminated arable soil (Deng et al., 2009). Resilience of nitrifying activity occurred over 1–2 years following contamination by Zn (Rusk et al., 2004; Ruyters et al., 2010a) and adaptation to the heavy metal resulted from a shift in the structure in the ammonia-oxidizing bacterial community that gradually dominated the nitrifying community (Ruyters et al., 2010b). Denitrifying activity, in contrast, was partly recovered 8 days after heavy-metal addition (Cu, Cd and Zn) with complete resilience after 2 months (Holtan-Hartwig et al., 2002). Exposure of cells extracted from the contaminated soils to heavy metals indicated that they had developed a tolerance to the heavy metals (Holtan-Hartwig et al., 2002), as has been shown for bacteria in general (Bååth et al., 1998, 2005; Fernández-Calviño et al., 2011). In contrast, no resilience of the denitrifying activity was observed 3 months after silver addition while silver induced the enrichment of novel denitrifiers (Throback et al., 2007).

Several factors can lead to the resilience of microbial communities and functions to heavy-metal contamination, including substitution of sensitive strains by tolerant ones; genetic modifications to produce heavy-metal resistance; transfer of genes encoding resistance or tolerance against heavy metals; or decreased heavy-metal bioavailability. The transfer of mobile genetic element by plasmids among taxonomically diverse bacteria in soil is well described (Springael & Top, 2004) and can contribute to the dissemination of genes that provide resistance to contaminant stress (reviewed in Smalla & Sobecky, 2002; Sobecky & Coombs, 2009). As such, the emergence of resistance to heavy metals or other contaminants in the soil microbial community can be regarded as a process describing deterioration of the ecosystems and is a bioindicator of contaminant exposition (Bérard et al., 2004).

Insights into the underlying mechanisms

Resistance, resilience and the biodiversity–ecosystem functioning relationships debate

According to the insurance hypothesis (Loreau et al., 2002), one of the proposed consequences of biodiversity loss is a reduction in the ecosystem stability. This hypothesis is based on the intuitive idea that the probability of finding species able to adapt to changing conditions and allowing ecosystem functioning is greater in a more diverse ecosystem. A test of this hypothesis in soil using differential fumigation with chloroform to decrease soil biodiversity showed reduced resilience of plant decomposition to heat and copper disturbance in soils with the lowest biodiversity (Griffiths et al., 2000). Subsequent experiments with an arable soil that was sterilized and then inoculated with diluted soil suspensions to alter biodiversity (Griffiths et al., 2001b) or with an upland grassland soil, whose biodiversity had been altered by both fumigation or dilution (Griffiths et al., 2004), showed no consistent effects of biodiversity on stability. This may have arisen because the way in which biodiversity is manipulated (i.e. fumigation or sterilization and inoculation) can affect the response of the soil microbial community to disturbance (De Ruiter et al., 2002). Thus, fumigation in particular may have selected for certain physiological traits in the surviving microorganisms (Griffiths et al., 2000, 2001b), while changes in stability are thought to be related to the traits of individual species in the community (Griffiths et al., 2004). An alternative approach is to experimentally build up different levels of biodiversity. Assemblages of up to 43 species of fungi did show evidence of increasing stability with increasing biodiversity (Setälä & McClean, 2004; Dang et al., 2005). As did communities containing up to 72 species of bacteria (Bell et al., 2005) in which there was a decelerating relationship between species richness and function (community respiration). This was consistent with there being functional redundancy between the species, but with the caveat that apparently unimportant species in a stable environment might have a role in maintaining function in a fluctuating environment, that is, be important for stability (Bell et al., 2005). However, soils contain far more species than generally used in such community assembly experiments and effects of biodiversity are more evident in systems with low diversity (Nielsen et al., 2011). Changes in species richness are most often considered when investigating the role of biodiversity for ecosystem functioning while biodiversity encompasses other components such as evenness (the relative abundance of species). Manipulation of both richness and evenness of the denitrifier community revealed that resistance to salinity disturbance was lower when initial communities were highly uneven or dominated by a few species, which demonstrated that not only richness but also of evenness can be of importance for ecosystem stability (Wittebolle et al., 2009).

Much research has also been conducted on the intermediate disturbance hypothesis (IDH). This hypothesis is based on the assumption that there is a trade-off between the ability to compete and the ability to withstand a disturbance. According to the IDH, intermediate levels of disturbance result in a higher biodiversity level because of the coexistence of organisms having different life strategies (i.e. r and K), which ensure ecosystem stability (Connell, 1978). Accordingly, resilience of methane oxidation activity to pyrene contamination was observed only at low pyrene concentration, which also increased biodiversity of the methane oxidizers. At high concentrations, diversity was decreased and no recovery was observed (Deng et al., 2011). Similarly, by examining soils from an experimental gradient of copper contamination, Wakelin et al. (2010b) showed that changes in microbial diversity followed a unimodal response (i.e. increased until a critical concentration then decreased) in line with the IDH.

Other studies have manipulated plant diversity to examine the effects on the resilience of soil microorganisms. Orwin & Wardle (2005) found no effect of plant biodiversity but a strong effect of plant species composition on the microbial resistance and resilience to a drought disturbance. They concluded that either nutrient limitation or soil microbial community structure may have altered soil resilience, while resistance was neither related to soil chemical nor to microbial community properties. Increasing the number of grassland plant species from one to six similarly did not alter soil functional stability (Griffiths et al., 2001a). On the other hand, Pfisterer & Schmid (2002) found an inverse relationship between plant species richness and microbial resistance and resilience to drought. Arbuscular-Mycorrhizal fungal communities proved high resilience to the removal of specific plant functional groups, showing a complete recovery in colonization after 30 months (Urcelay et al., 2009), but it was speculated that ericoid and ectomycorrhizal fungi would be less resilient to changes in vegetation. Community composition and richness of arbuscular-mycorrhizal fungal communities in Plantago lanceolata roots were also found to be resistant to disturbance, which was not because of a recolonization of the disturbed area (Lekberg et al., 2011),

These studies showed that the influence of biodiversity on ecosystem stability is complex and depends not only on species richness but also on the evenness or composition of the soil microbial community. Resistance and resilience to disturbance might also vary between functional microbial microbial guilds dependent on their levels of functional redundancy. A high level of functional redundancy, within a functional community, that is, a high number species performing the same function, might act as a buffer against the effect of biodiversity loss on functioning. For example, manipulation by dilution of microbial biodiversity showed that resistance and resilience to a heat disturbance differed between the microbial communities studied, with denitrifiers being less affected than nitrite oxidizers (Wertz et al., 2007). As denitrifiers are more diverse than nitrite oxidizers, it was hypothesized that functional redundancy was higher for denitrifiers that buffered the effect of their decreased diversity. Functional redundancy is also supported by the finding that a narrow scale function (decomposition of dichlorophenol) was significantly reduced following benzene or copper addition, whereas a broadscale function (decomposition of wheat shoots) was unaffected (Girvan et al., 2005). However, the contribution of the redundancy of the soil microbial community to resilience might not hold for specific soil functions if the microbial community is already reduced in diversity because of some previous disturbance (Liebich et al., 2007). In any case, redundancy is a multifaceted concept and whether it really exist or not in natural ecosystem is still debated (Loreau, 2004). Studies addressing the role of redundancy in microbial ecology are also hampered by the difficulty to accurately define functional categories and limited knowledge of the true diversity of the corresponding microbial guilds. For example, functional redundancy was assumed to be low among microorganisms performing the first step of nitrification until the finding in 2005 that microorganisms belonging to another domain, the crenarchaea were also capable to oxidize ammonium into nitrate (Könneke et al., 2005). Finally, the apparent inconsistency of data on the role of microbial diversity for soil functioning is not surprising because the effect of the soil microbial diversity is interwoven with many other factors such as: interactions between species; soil properties; or the disturbance history as discussed below.

Ecological networks

Together with diversity, the importance of interactions between species for ecosystem stability has long been studied in ecology using ecological network to describe these interactions (May, 1972; Pimm, 1984; Montoya et al., 2006). Early theory paradoxally predicted that more complex network are likely to be less stable (May, 1974), the loss of one species following disturbance leading more easily to secondary extinction in highly connected networks (Solé & Montoya, 2001; Dunne et al., 2002). In contrast, compartmentalization, that is, the existence of groups of species that have a higher probability of interacting with each another than with other species, significantly increases both resistance and resilience against perturbation because compartments act to buffer the propagation of extinctions (May, 1972; Stouffer & Bascompte, 2011). The relationship between network architecture and stability can also be affected by the type of interaction. A highly connected and nested architecture promotes community stability in mutualistic networks, whereas stability is increased in compartmented and weakly connected architectures in trophic networks (Thébault & Fontaine, 2010). Allesina & Tang (2012) showed that weak interactions can be either stabilizing or destabilizing depending on the type of interactions between species. Here, weak interactions and a realistic food web structure (as opposed to unstructured networks in which species interact at random) were found to decrease the stability of predator-prey systems. Despite a large body of literature, the identification of the network properties involved in ecosystem stability is still ongoing (Montoya et al., 2006).

Ecological networks are commonly used to understand the resistance of native communities to invasion by new species (Romanuk et al., 2009), which can be considered as a disturbance. While natural invasion by exotic species is difficult to study in microbiology, there is a large body of literature investigating the effects of the introduction of new microbial species in soil, mostly as pest biocontrol, on indigenous microorganisms. In most cases, introduction of a fungal or bacterial biocontrol agent had a minimal impact on the soil microbial community composition. Indeed, application of the nonpathogenic fungal strains of Fusarium oxysporum or Trichoderma atroviride showed minor shifts in both the bacterial and fungal communities that lasted only a few weeks (Edel-Hermann et al., 2009; Savazzini et al., 2009). Similarly, bacterial biocontrol agents caused only minor and transient modifications of the microbial community composition (Bankhead et al., 2004; Scherwinski et al., 2007; Correa et al., 2009). Altogether these results suggest a strong resiliance and resistance of the native microbial community to invading microorganisms. However, a few studies also reported a significant impact of microbial inoculants on the bacterial community structure (Kozdroj et al., 2004). This could be explained by recent work indicating that, accordingly to ecosystem theory, susceptibility of soil ecosystems to invading microbial species is dependent on their complexity (Fließbach et al., 2009).

Tolerance and adaptation of soil microorganisms

Response of individual cells to disturbance, which has consequences for the stability of the total community, is related to the activation of protective or adaptative mechanisms for surviving. Transcriptional regulation of the genes whose products are involved in physiological tolerance and adaptation to withstand disturbances has been described elsewhere (Ramos et al., 2001, 2009). Physiological mechanisms on the effects of drought and dry/wet cycles were reviewed by Schimel et al. (2007). Exposure to drought results in an accumulation of osmolytes by microorganisms either by producing organic solutes or by taking up ions from the extracellular solution (Csonka, 1989) to maintain cell integrity. Rehydration after a long period of drought can be as stressful as dehydration itself and to prevent a rupture of the cell walls during soil rewetting, microorganisms release the osmolyte carbon resulting in a soil respiration flush (Fierer & Schimel, 2003). In the short term, changes in microbial physiology and the energetic cost of adaptive mechanisms for withstanding a disturbance can affect the resistance and resilience of microbial processes. Thus, physiological effects associated with cell dehydration rather than substrate diffusion limitation were inhibiting nitrification at water potentials lower than −0.6 MPa (Stark & Firestone, 1995). Because the energetic cost of adaptation can differ between microorganisms, in the long–term, adaptation can also result in microbial community composition shifts (Schimel et al., 2007).

Several studies have addressed the effect of heat disturbance on microorganisms and the related molecular mechanisms, which include protein denaturation with disruption and possible destruction of both secondary and tertiary structures. Resistance and adaptation of microorganisms to increased temperature are most often owing to the synthesis of heat shock protein folding and unfolding other proteins (Ramos et al., 2001; Tobor-Kaplon et al., 2006b). Interestingly, induction of heat shock proteins is triggered by exposure to other environmental stressors such as osmotic shock or the presence of heavy metals and aromatic compounds (Ramos et al., 2001), which provides a molecular basis for cross-protection (where exposure to one disturbance increases resistance to a different disturbance, Ventura et al., 2006). The ability of some bacteria to form thick-walled and highly resistant spores is also an efficient way to cope with a heat disturbance but also to protect the bacteria from a large range of other environmental stressors. The modifications induced in the soil microbial community by relatively minor disturbances, such as copper, heat or atrazine, led to a significantly increased resistance of the community structure to a subsequent severe disturbance in the form of mercury contamination (Bressan et al., 2008).

Bacterial cells can acquire resistance to xenobiotic compounds through the transfer of genes or genetic mobile elements. There is a large body of evidence of the transfer in soil of organic xenobiotic-degrading genes (reviewed in Springael & Top, 2004). For example, enhanced biodegradation of the herbicide atrazine, which is used to control weeds in maize production, after repeated herbicide applications has been attributed to horizontal transfer of atrazine degrading genes (Devers et al., 2005, 2007). Mobile genetic element and horizontal gene transfer are important for adaptation not only by disseminating the existing xenobiotic-degrading pathways but also by constructing new ones. Thus, assembly of pre-existing gene motifs to generate new genes and recruitment of genes from other pathways combined with mutation events can lead to new pathways (Copley, 2000). Ability to assemble together novel degradation genes has been shown in vitro by construction of a pesticide degrading gene after DNA shuffling (Boubakri et al., 2006). Adaptation to organic xenobiotics by acquisition of the corresponding degrading genes through horizontal gene transfer and patchwork assembly can lead to changes in microbial communities because of the competitive advantage conferred by the ability to incorporate these compounds into their diets (Copley, 2000; Springael & Top, 2004; Ramos et al., 2009; Udiković-Kolić et al., 2011). Such shifts in microbial diversity are in turn likely to affect the resistance and resilience of soil microorganisms to subsequent disturbances.

Contrasting mechanisms have been proposed to account for tolerance to disturbance. Thus, there is a general increase in the physiological tolerance of the microorganisms in contaminated habitats, favouring generalist microorganisms (Atlas et al., 1991), but an alternative mechanism may occur in specifically metal contaminated sites. Here, individual microorganisms show an increased tolerance by only using high energy substrates despite a reduction of catabolic versatility (Wenderoth & Reber, 1999; Witter et al., 2000). Finally, as discussed previously, exposure of the soil microbial community to disturbance can lead to tolerance and adaption of individual cells to subsequent disturbances.

Effects of previous disturbance

To what extent exposure to an initial disturbance is able to affect the stability of soil microorganisms to subsequent disturbances is still unclear. Already stressed microbial communities can be more or less stable to a second disturbance depending on the nature of the disturbance (Tobor-Kaplon et al., 2005). In later work, Tobor-Kaplon et al. (2006a) found that respiration in the most metal contaminated soils was more affected by a subsequent heat or salt disturbance, which supports the hypothesis that stressed systems have less energy to cope with additional disturbance than previously undisturbed systems. Thus, organisms from highly polluted soils have lowered resources because of the allocation of energy to detoxification and damage repair caused by the first disturbance, which makes any additional disturbance harder to cope with (Calow, 1991; Kuperman & Carreiro, 1997). Lower stability after a new disturbance could also be due to decreased diversity following exposure to an initial disturbance. Indeed, Müller et al. (2002) suggested that the decreased resistance in a heavy-metal contaminated soil as compared to the control was because of a reduced microbial diversity. However, measurements of bacterial growth rates in the gradient of metal polluted soils showed that the least contaminated ones were the least stable to increased temperature, suggesting that disturbed systems can also be more stable because they can gained abilities (adaptation and physiological changes) to cope with additional disturbance (Tobor-Kaplon et al. (2006b). Similarly, hydrocarbon polluted soils exhibited greater resilience of plant decomposition to experimentally applied heat and copper disturbances than unpolluted control soils (Griffiths et al., 2001a). Increased physiological tolerance of microorganisms in contaminated habitats (Atlas et al., 1991) has also been discussed earlier. One could also argue that whether the mechanisms with which the organisms deal with the initial and additional disturbances are related or not will likely to contribute to the microbial community stability to successive disturbances. Thus, primary disturbance because of copper addition, but not to heating or pesticide addition, increased the subsequent resilience of the soil nitrate reducing activity to another heavy metal, mercury, which suggests that the relatedness of the disturbances influences the outcome (Philippot et al., 2008). However, soils with a history of copper-contaminated sludge were not more resistant and resilient to laboratory applied copper disturbance than soils amended with uncontaminated sludge (Griffiths et al., 2005). Mertens et al. (2007) also showed that the stability of nitrification to disturbances (biocide, freeze–thaw and dry–wet) was not related to previous Zn contamination and depended on land use history and the nature of the disturbance.

Tobor-Kaplon et al. (2006a, b) and van der Wurff et al. (2007) suggested that the responses of microbial processes and communities to disturbance depend on the nature of the disturbance and whether a subsequent disturbance is similar to the first, in terms of the mechanisms with which the organisms deal with the disturbance.

Role of soil properties

A clear connection between stability and soil structure was demonstrated by reduced resistance after grinding of the soil, which destroyed its structure. Both resistance and resilience were reduced in revegetated degraded soils when the soil structure had been destroyed (Zhang et al., 2010). Similarly, resistance of cellulose decomposition to chemical disinfection was decreased when the soil was ground to destroy structure (Fujino et al., 2008). It is likely that highly structured pore networks provide a shelter for soil microorganisms, akin to the protected pore space that shelters bacteria from faunal predation in soil (Heijnen & Van Veen, 1991). A survey of 26 soils across Scotland showed that organic carbon content was correlated to resilience after chemical (Cu addition) and physical (compaction) disturbances, whereas resilience after heat was correlated to none of the soil properties. This study also revealed that there was no soil uniformly resilient to all disturbances, so that soils resistant to one disturbance tended to be susceptible to a different disturbance (Kuan et al., 2007). When this data were combined with geographic information system (GIS) techniques and the national soil data base, multiobjective regression tree analysis was able to produce a national map of soil resistance and resilience characteristics (Debeljak et al., 2009) (Fig. 3). Resilience of soils to Triclosan (a broad-spectrum antimicrobial agent used in healthcare products) also correlated with organic matter and clay content (Butler et al., 2011). It was speculated that soil properties affected bioavailability of the Triclosan although differences in resilience might also reflect differences in microbial community structure. Confounding effect of soil properties and diversity were also observed by Girvan et al. (2005). Thus, a greater resistance of the genetic diversity and functional resilience to benzene were found in a more diverse organo-mineral soil than in a less diverse mineral soil.

Figure 3

Risk based maps of Scotland, such as this of overall soil stability (resistence and resilience) determined from the responses to four disturbances (copper, heat, compression and waterlogging) from Debeljak et al., 2009, may be a useful addition to aid decision makers.

Because the diversity (or composition) of the microbial community present in a soil is strongly dependent on its physico-chemical properties, it is very difficult to distinguish the importance of these two factors for soil resistance and resilience. To disentangle the contribution of abiotic and biotic soil factors to soil stability, Griffiths et al. (2008b) inoculated 26 different sterile soils with a single organism, Pseudomonas fluorescens. They showed that the resistance to copper or heat disturbance varied depending on the soils. The resilience of this organism also varied when inoculated in two contrasting, sterile soils (Griffiths et al., 2008b). Similarly, Bárcenas-Moreno et al. (2011) concluded that community assembly when inoculating sterile soil was driven by soil pH. These results demonstrate the importance of the soil physico-chemical properties for its stability. A transplant experimentation was further designed to swap microbial communities between the two contrasting soils by inoculating two different communities into each of two sterile soils. Interestingly, while in one soil, the different microbial communities inoculated converged resulting in a similar microbial community structure, patterns of resilience remained soil specific (Griffiths et al., 2008b).

Concluding remarks

Multiple factors influence soil stability (resistance and resilience). It is clear that it is related to soil properties such as organic matter, aggregation, the quantity and quality of carbon inputs and, to a lesser extent, clay content and soil pH. As a consequence, land management can increase (e.g. through the addition of organic residues to soil) or decrease (e.g. through the tillage of grassland or conversion of forest to agriculture) soil stability. The role of microbial diversity in soil stability is not simply linked to the absolute number of species present but is related to the functional traits of those species, although there appear to be conflicting conclusions. While on one hand microbial community structure is not a major factor and that stability is independent of microbial biomass, on the other hand, the soil response to disturbance depends on microbial cell characteristics and specific microbial functions (such as increased metabolic versatility in contaminated soils, physiological responses and adaptation). However, these conclusions are compatible given the poor resolution of microbial community diversity analyses, which describe only the dominant populations. We propose that soil biological stability is governed by the physico-chemical structure of the soil through its effect on microbial community composition and microbial physiology (Griffiths et al., 2008b) and that there is no general soil response to disturbance because stability is particular to the disturbance and soil history.

As soil stability results from a combination of soil physico-chemical characteristics and species level characteristics of the microbial community, it bridges these two domains. This could provide a quantitative measure of soil health, which could translate into policy advice and improved land management practices (Griffiths et al., 2001a; Kibblewhite et al., 2008). Stability measurements could also provide an indirect indication of ‘critical slowing down’ and the approach of ecosystem tipping points (Veraart et al., 2011). But rather than studying the effects of experimental disturbances, monitoring soil responses to stochastic natural disturbances may be a useful early indicator of an impending ecological state change (Van Nes & Scheffer, 2007).


The authors would like to thank the French Embassy in Dublin for supporting the collaboration between the INRA and Teagasc, and Dr Ayme Spor for helpful comments. B.S. Griffiths acknowledges support from Science Foundation Ireland under their Stokes Professorship initiative; Grant No. 07/SK/B1236b. This work was also partly supported by the European Commission within the EcoFINDERS project (FP7-264465).


  • Editor: Eva Top


View Abstract