Effect of primary mild stresses on resilience and resistance of the nitrate reducer community to a subsequent severe stress


  • Present address: Mickael Cregut, UMR INPL-ENSAIA-INRA, Agronomie et Environnement Nancy-Colmar, 2 avenue de la forêt de Haye, BP 172, 54500 Vandoeuvre Les Nancy, France.
    Mélanie Bressan, CEA/Cadarache, DSV-DEVM, UMR 6191 CNRS-CEA- Université de la Méditerranée, Laboratoire d'Ecologie Microbienne de la Rhizosphère, 13108 Saint-Paul-Lez-Durance, France.

  • Editor: Elizabeth Baggs

Correspondence: Laurent Philippot, INRA, University of Burgundy, UMR 1229 Soil and Environmental Microbiology, CMSE, 17, rue Sully, B.P. 86510, 21065 Dijon Cedex, France. Tel.: +33 3 80 69 33 46; fax: +33 3 80 69 32 24; e-mail: philippo@dijon.inra.fr


The factors regulating soil microbial stability (e.g. resistance and resilience) are poorly understood, even though microorganisms are essential for ecosystem functioning. In this study, we tested whether a functional microbial community subjected to different primary mild stresses was equally resistant or resilient to a subsequent severe stress. The nitrate reducers were selected as model community and analysed in terms of nitrate reduction rates and genetic structure by narG PCR-restriction fragment length polymorphism fingerprinting. Heat, copper and atrazine were used as primary stresses and mercury at a high concentration as a severe stress. None of the primary stresses had any significant impact on the nitrate reducer community. Although primary stress with heat, copper or atrazine had no effect on the resilience of the nitrate reducer activity to mercury stress, pre-exposure to copper, another heavy metal, resulted in increased resilience. In contrast, the resistance of both structure and activity of the nitrate reducer community to severe mercury stress was not affected by any of the primary stresses tested. Our experiment suggests that the hypothetical effect of an initial stress on the response of a microbial community to an additional stress is complex and may depend on the relatedness of the two consecutive stresses and the development of positive cotolerance.


Soil ecosystems are bound to experience natural or anthropogenic stresses. The capacity of the soil to withstand a stress is known as resistance whereas its capacity to recover after a stress is referred to as resilience (Pimm, 1984). Owing to the increased inputs of toxicants such as heavy metals or pesticides in agriculture, the response of soil microbial communities to these toxicants, which constitute stresses for soil organisms, has been investigated in several studies (Frostegård et al., 1993; Witter et al., 1994; Díaz-Raviña & Bååth, 1996; Pell et al., 1998; Kelly et al., 1999; Ranjard et al., 2006a, b). Frostegård et al. (1996), for example, reported changes in microbial community structure, as determined by phospholipid fatty acid analysis, in soils contaminated with zinc. Similar observations were also made by Griffiths et al. (1997) and Bååth et al. (1998) in soils contaminated by other heavy metals. Other changes such as decreased microbial biomass, growth or respiration have been reported as being due to heavy metal contamination (Giller et al., 1998). The influence of other stresses such as pesticide addition on the structure or the activity of microbial communities has also been widely investigated and early studies reported inhibitory effects of pesticides on microbial activity such as denitrification (Bollag & Barabasz, 1979).

To what extent exposure to an initial stress is able to affect the response of soil microorganisms, in terms of resistance or resilience, to an additional stress is still a matter of debate. Odum (1981) hypothesized that stressed ecosystems are more resistant due to adaptative changes to withstand stress. Most research to date has focused on above-ground communities such as plants while a few studies have been performed to verify these theories using soil microorganisms, which play a key role in soil functioning. Tobor-Kaplan et al. (2006) studied the effect of an additional stress in the form of lead or salt on soil microbial communities that had been exposed for more than 20 years to different levels of copper or low pH. They showed that the effect of additional stress depended on the type of stress and that highest resistance and/or resilience occurred in the least contaminated soils. However, these studies were conducted at the total microbial community level whereas it has been suggested that the impact of stress is less pronounced at the functional group level than at the species level (Schindler, 1990; Vinebrooke et al., 2003).

The objective of this study was to determine whether exposure of a functional bacterial community to a primary mild stress would affect its resistance or resilience to a subsequent severe stress. For this purpose, we used the nitrate-reducing bacteria involved in the nitrogen cycle as a model functional community because they constitute a taxonomically diverse group including members of the Alpha-, Beta-, Gamma-, Epsilonproteobacteria, high and low GC Gram-positive bacteria and are widespread in the environment (Philippot, 2005; Philippot & Hallin, 2005). Three different types of primary mild stresses were applied to the soil for 2 weeks. After 1 month, the stressed and control soils were exposed to severe stress and the genetic structure and activity of the nitrate reducer community were monitored for 3 months. Metal contaminations at a low (copper) and a high concentration (mercury) were used as mild and severe stress, respectively. The two other mild stresses applied to the soil microcosms were heat shock and herbicide (atrazine) addition.

Materials and methods

Soil microcosms

The soil used in this study was collected from a Eutric cambisol in an INRA experimental field located 10 km from Dijon. It consisted of: clay 43.8%, sand 5.9%, silt 50.3%, organic carbon 1.5% and total nitrogen 0.16%. After sieving at 2 mm and homogenization, the soil sample was divided into 50 g aliquots and randomly distributed into 144 jars with three replicate microcosms per treatment and per sampling date. The treatments were: (1) untreated (control), (2) addition of CuCl2 (50 g g−1 dry soil), (3) addition of the herbicide atrazine (1.5 g g−1 dry soil) and (4) heating (35 °C for 48 h). These primary stresses (CuCl2, atrazine and heat) were applied at days 0 and 15. At day 30, the severe stress [addition of HgCl2 (100 μg g−1 dry soil)] was applied to half of the microcosms of each treatment. The microcosms were then incubated at 20 °C and watered regularly with sterile distilled H2O to maintain a constant humidity of 80% WHC. Three replicate microcrosms per treatment were analysed after 0, 15, 30, 45, 60, 90 and 120 days.

Measurement of potential nitrate reductase activity

The potential nitrate reductase activity was determined by anaerobic incubation of the soil following a modified protocol of Kandeler (1995). Briefly, 0.2 g of soil was weighed into four equal replicates and incubated in a total volume of 1 mL containing 1 mM potassium nitrate with 206 μL of 2,4-d-dinitrophenol added to inhibit nitrite reduction. After a 24-h incubation at 28 °C, the soil mixture was extracted with 4 M KCl and centrifuged for 1 min at 13 000 g. The accumulated nitrite in the supernatant was determined by a colorimetric reaction using a reagent composed of N1-naphthyl ethylenediamine dichloride (2 g L−1) and sulphamide (40 g L−1) (Kandeler, 1995).

Genetic structure analysis of the nitrate-reducing community

The genetic structure of the nitrate reducer community was assessed using the narG gene encoding the membrane-bound nitrate reductase as a molecular marker (Philippot et al., 2002). DNA was extracted from 1 g of soil samples from the 144 microcosms as described previously (Martin-Laurent et al., 2001; Ranjard et al., 2003). Extracted DNA was quantified by spectrophotometry at 260 nm using a BioPhotometer (Eppendorf, Hamburg, Germany). Amplification was carried out on 10 ng of purified soil DNA using the narG primers according to Philippot et al. (2002). Three independent PCRs were performed on each sample. The narG PCR products were purified using the MiniElute gel extraction kit (Qiagen, France). The purified PCR products were digested with AluI restriction enzyme at 37 °C for 12 h and the narG restriction fragment length polymorphism (RFLP) fingerprints were obtained after separation by electrophoresis on a native 6% acrylamide gel for 11 h at 5 mA in TBE buffer.

Statistical analysis of the data

Significant differences (P<0.05) in nitrate reductase activity between treatments at each date were determined using statview-se software and Student's t-test (n=3). The resistance of the nitrate reduction activity (NRA) to mercury stress was estimated by calculating the percentage of changes from the uncontaminated soil for each treatment:


The narG fingerprint gels were analysed using the 1d-scan software (ScienceTec, Les Ullis, France) and data were converted into a table summarizing the band presence (i.e. peak) and intensity (i.e. height or area of peak) using the preprisa program. Principal component analysis (PCA) was performed on the data matrix using the ade-4 software (Thioulouse et al., 1997). Statistical ellipses were drawn over the replicates of a treatment and represent 90% confidence.

Results and discussion

Effect of primary mild stresses on the structure and activity of nitrate reducers

The primary stresses e.g. heat, addition of either atrazine or copper at agronomical concentrations, were chosen because of their agronomical relevance. Stresses such as heat shock or copper addition have also been selected previously in other studies to investigate the resilience of the soil microbial community (Griffiths et al., 2001; Girvan et al., 2005; Tobor-Kaplan et al., 2006). No significant differences in NRA were observed between treatments after exposure to primary stresses for one or two weeks (Fig. 1). Similarly, PCA of the narG fingerprints from the 144 microcosms did not show any difference in the structure of the nitrate reducer community between the different microcosms at days 0, 15 (data not shown) or 30 (Fig. 2). Therefore, none of the primary mild stresses had any detectable impact on the structure or activity of the nitrate reducer community, before the mercury stress. Several studies, consistent with ours, showed that nitrate reducers or denitrifiers were not affected by atrazine treatments at concentrations ranging from 5 to 100 μg g−1 soil (Yeomans & Bremner, 1987; Pell et al., 1998; Philippot et al., 2006). The absence of a copper effect was also reported by Griffiths et al. (2004) and Girvan et al. (2005), who found that copper had no or little effect on microbial community structure over a short incubation period. In contrast, slight modifications of the microbial community structure in response to copper addition were observed by Ranjard et al. (2006a, b). However, these modifications were transient. The absence of an effect of heat stress on nitrate reducers is also consistent with previous studies showing that heating the soil at 40 °C for 18 h did not modify bacterial respiration in a grassland soil (Griffiths et al., 2001).

Figure 1.

 Nitrate reduction rates in μg N–NO3 reduced g−1 dry soil day−1 in microcosms contaminated by mercury (at day 30) or in uncontaminated microcosms after initial exposure to a primary stress (control, atrazine, heat choc or copper) from days 0 to 120. * Indicates significant differences between mercury-contaminated microcosms at each date (P<0.05). Scale bars are SEs (n=3).

Figure 2.

 PCA ordination of the nitrate reducer genetic structure for control microcosms (a) and microcosms pre-exposed to atrazine (b), heat (c), or copper (d) and contaminated (grey ellipses) or uncontaminated by mercury (white ellipses) from days 30 to 120. The statistical ellipses indicate 90% confidence for replicates.

Impact of mercury addition on nitrate reducers

Application of a high concentration of mercury resulted in an enhancement of NRA in all treatments to c. 35 μg NO3 g−1 soil day−1 at day 45 (Fig. 1), resulting in a three- to four-fold increase within 2 weeks compared with the uncontaminated microcosms (Fig. 3). This stimulation was surprising because high concentrations of heavy metals are known to influence the activity of soil microbial communities by altering the conformation of enzymes, blocking essential functional groups or by exchanging metal ions (Tyler, 1981). Kandeler et al. (1996) concluded that heavy metals had a more negative influence on enzymatic processes in the nitrogen cycle than on those in the carbon cycle. Indeed, an inhibition of denitrification rates after addition of heavy metals to soils was observed in numerous studies (Bardgett et al., 1994; Holtan-Hartwig et al., 2002). However, a few studies reported a stimulation of enzyme activities after the addition of heavy metals. For example, ammonification, nitrification and denitrification activities were enhanced by addition of a low concentration of heavy metals (Yang et al., 2005). Griffiths & Hallett (2005) found that the addition of zinc or copper at concentrations higher than 50 mg kg−1 increased basal respiration in soil. The stimulation of the NRA observed in our study could be due to (1) death of organisms caused by the addition of mercury, which then provided a readily available source of carbon for nitrate reduction, and/or (2) a general response to stress causing an unspecific increase of microbial activity. The first hypothesis is supported by the slow decrease in the nitrate reduction rate observed in the soils without mercury, i.e. from 9.1 μg N–NO3 g−1 soil day−1 at day 0 to 4.2 μg N–NO3 g−1 soil day−1 at day 120, which suggests that carbon may have been limiting in our microcosms.

Figure 3.

 Percentage of changes in the NRA between the mercury-contaminated soils and the uncontaminated soils in the microcosms exposed to a primary stress (atrazine, heat or copper) and the control microcosms. Scale bars are SEs (n=3).

The stimulating effect of mercury addition on the NRA was transient. After day 60, the nitrate reduction rates diminished considerably in the contaminated soils and a partial or even a total restoration of the original NRA was observed at day 120 (Fig. 3). This recovery of NRA could result (1) from depletion of the carbon released from dead organisms after mercury addition or (2) from metal ageing leading to a decrease in mercury biovailability. In contrast, Throbäck et al. (2007) and Griffiths et al. (2000) found no significant recovery of respiration from a negative effect of heavy metal application and the latter suggested that heavy metals exert a strong persistent stress that prevents resilience. On the other hand, a full recovery of respiration and potential denitrification activity after exposure to heavy metals was observed by Holtan-Hartwig et al. (2002) and Tobor-Kaplon et al. (2005), respectively. The impact of heavy metals on denitrifiers is likely to be dependent on the soil type and on the concentration and nature of the heavy metals (Philippot et al., 2007), which may explain these contrasting results.

Although an obvious effect of mercury addition on nitrate reducer activity was observed after 2 weeks, no clear effect on the genetic structure of the nitrate reducer community was observed for most treatments before 120 days as shown by the absence of discrimination on principal axis 1 of the PCA (Fig. 2). However, a clear effect of mercury addition was observed in all treatments only at day 120 (Figs 2 and 4). This indicates that the increased rates of nitrate reduction at day 45 were not correlated with changes in dominant populations within the nitrate reducer community. Difficulties in linking the structure and activity of a functional community by DNA fingerprinting have already been highlighted (Philippot & Hallin, 2005). The changes in nitrate reducer community structure observed after 2–3 months are likely to be due to the replacement of mercury-sensitive populations of nitrate reducers by ones more tolerant to this stress. Our results are consistent with those of previous studies that showed shifts in the total microbial community structure in soils contaminated with heavy metal compared with the corresponding noncontaminated soil (Frostegård et al., 1996; Griffiths et al., 1997; Ranjard et al., 2006a, b).

Figure 4.

 Fingerprinting analysis of the genetic structure of the nitrate reducer community in control microcosms and microcosms pre-exposed to a primary stress (atrazine, heat or copper) and contaminated or uncontaminated by mercury at day 120. The molecular size marker on both sides was VIII from Boehringer Mannheim.

Effect of primary stress on the resistance and resilience of nitrate reducers to a subsequent stress

We did not detect any significant differences in the maximum percentage change in NRA between treatments, indicating that, in our experiment, exposure to primary stresses did not affect the resistance of NRA to mercury stress (Fig. 3). However, comparison of the nitrate reduction rates between treatments at days 90 and 120 revealed an effect of the primary stresses on resilience. Thus, at day 90, the effect of mercury on NRA decreased significantly in the copper-pre-treated microcosms but not in the control, heat- and atrazine-pretreated microcosms (Fig. 3). In addition, at day 120, no significant differences were observed in the nitrate reduction rates between mercury-contaminated and uncontaminated microcosms for the copper pre-treatment. In contrast, for the control, heat and atrazine-treatments, the nitrate reduction rates still remained twofold higher in the microcosms exposed to mercury compared with the uncontaminated microcosms.

This more rapid recovery of NRA in the soil pre-exposed to copper than in the control soil suggests that pre-exposure to this metal resulted in higher resilience to metal contamination (Figs 1 and 3). Similarly, Griffiths & Hallett (2005) showed that contamination with zinc increased resilience to copper stress. An increase in tolerance to metals other than the metal originally added to soil was also observed by Díaz-Raviña & Bååth (1996), indicating the existence of multiple heavy-metal tolerance at the community level. It was suggested that this increased tolerance was due to physiological and/or genetic adaptation (Díaz-Raviña & Bååth, 1996). This is supported by the fact that the mechanism of toxicity is very similar for the heavy metals and consists mainly in the disruption of essential biological molecules, such as protein, enzyme and DNA. Here, we can hypothesize that the positive correlation between species tolerance to heavy metals increases the resistance of nitrate reducers to mercury stress as a result of exposure to copper. Therefore, our results suggest that initial exposure to the primary stresses affected the resilience of the NRA to an additional stress but only when initial and subsequent stresses were of the same nature.

In contrast, in all treatments, exposure of the microcosms to mercury resulted mainly in changes in the intensity of the same bands, suggesting that the same populations were selected whatever the primary stress applied (Fig. 4). Altogether, the fingerprinting analysis suggests that the minor stresses applied did not influence the resistance of the nitrate reducer community structure. Unfortunately, because clear differences between mercury-contaminated and uncontaminated soils were not observed until day 120, the incubation period of our experiment was too short to test whether the primary stresses had an effect on the resilience of the genetic structure of the nitrate reducer community. However, Griffiths et al. (1997) observed that the resilience of microbial community structure to heavy metal stress could be very low so that differences in microbial community composition between soils contaminated with heavy metals and uncontaminated soil were still observed almost 3 years after contamination.


This study shows that initial exposure of soil to a primary stress can modify the stability of a functional community to a subsequent stress. Thus, a higher resilience of nitrate reducer activity to mercury stress was observed in microcosms pre-exposed to copper, another heavy metal. However, no significant effect of copper on response of the genetic structure of the nitrate reducers to mercury contamination was observed, indicating that the activity and structure of a given functional community might differ in stability. Our experiment suggests that the effect of an initial stress on the response of the microbial community to an additional stress is complex and will depend on the relatedness of the two consecutive stresses and on the development of positive cotolerance. Thus, the persistence of traits that determine the tolerance to one stress can influence the response of a functional community to other stresses.


The authors would like to thank S. Hallet for her technical assistance. This work was supported by a grant from the INRA ACI Microbiologie ‘Resilience of microbial ecosystems’.