LETTERS PUBLISHED ONLINE: 29 JANUARY 2012 | DOI: 10.1038/NCLIMATE1368 Land use alters the resistance and resilience of soil food webs to drought Franciska T. de Vries1 *, Mira E. Liiri2 , Lisa Bjørnlund3 , Matthew A. Bowker4 , Søren Christensen3 , Heikki M. Setälä2 and Richard D. Bardgett1 Soils deliver several ecosystem services including carbon sequestration and nutrient cycling, which are of central importance to climate mitigation and sustainable food production1–3 . Soil biota play an important role in carbon and nitrogen cycling, and, although the effects of land use on soil food webs are well documented4–6 , the consequences for their resistance and resilience to climate change are not known. We compared the resistance and resilience to drought—which is predicted to increase under climate change2,7 —of soil food webs of two common land-use systems: intensively managed wheat with a bacterial-based soil food web and extensively managed grassland with a fungal-based soil food web. We found that the fungal-based food web, and the processes of C and N loss it governs, of grassland soil was more resistant, although not resilient, and better able to adapt to drought than the bacterial-based food web of wheat soil. Structural equation modelling revealed that fungal-based soil food webs and greater microbial evenness mitigated C and N loss. Our findings show that land use strongly affects the resistance and resilience of soil food webs to climate change, and that extensively managed grassland promotes more resistant, and adaptable, fungal-based soil food webs. Soils deliver a range of ecosystem services. They not only provide nutrients for crop growth, but also store significant quantities of C and N, which contributes to climate mitigation8 and sustainable food production1–3 . The soil organisms that drive processes of C and N cycling are significantly affected by land use and the intensity of agricultural management9 : not only can agricultural intensification reduce the diversity of soil biota4,6 , but it can also induce shifts in the composition of soil communities from fungal- to bacterial-based food webs4–6 . It has been suggested that fungal-based soil food webs, which are common in extensively managed farming systems9,10 , are better able to withstand climate change-related disturbances than are bacterial-based soil food webs11,12 . Moreover, because fungal-based soil food webs are linked to increased soil C and N retention5,13–17 , it is likely that they will also better retain C and N under climate-change-related disturbances than bacterial-based soil food webs. However, it has also been suggested that fungalbased soil food webs would recover slowly after climate changerelated disturbances, because slow-growing organisms, including fungi and fungal-feeding fauna, tend to be more resistant (that is, have greater ability to withstand a disturbance)18 , but less resilient (that is, have lower rate of recovery after a disturbance)18 , than fast-growing organisms such as bacteria and bacterial-feeding fauna11,18–20 . So far these ideas remain untested, and therefore the consequences of land use for the resistance and resilience of soil food webs and processes of C and N loss to climate-induced disturbances, such as drought, are not known. Here, we tested how contrasting agricultural land use affects the resistance and resilience of soil food webs and soil C and N loss to drought, which is predicted to increase under climate change2,7 . Specifically, we tested the hypothesis that fungal-based soil food webs, and the processes of C and N loss that they govern, are more resistant to drought than their bacterial-based counterparts. To test this hypothesis, we used two land use systems, namely extensively managed grassland with a fungal-based food web and an intensively managed wheat system with a bacterial-based food web. Effects of prolonged drought on soil food webs, including the microbial community, protozoa, nematodes, enchytraeids and microarthropods, were first tested in the field in both land-use systems (Methods and Supplementary Methods). Subsequently, the resistance and resilience of these soil food webs to simulated drought was tested using a laboratory assay on soils taken from the field experiment (Methods). Here, soil-food-web responses were measured at the end of the laboratory-based drought, and 1, 3, 10 and 77 days after rewetting. We constructed a soil-food-web model (Supplementary Fig. S1) to calculate total biomasses of the fungal and bacterial energy channels21 , and we used evenness of microbial phospholipid fatty acids (PLFAs) as a measure of changes in microbial communities (Supplementary Methods). We also measured key processes that are carried out by the soil food web and represent C and N loss pathways: CO2 and N2 O production, and N leaching in drainage waters. To quantify the immediate effect of drought on the soil food web and C and N losses, and their rate of recovery after rewetting, resistance and resilience indices were calculated22 . The resistance index calculates the absolute amount of change relative to the control at the end of the disturbance, and ranges from −1 to +1, with +1 indicating maximum resistance (that is, no effect of drought). The resilience index calculates the absolute difference that exists between drought and control treatments relative to the initial absolute effect of the disturbance, and ranges from −1 to +1, with +1 indicating maximum resilience (that is, complete recovery after rewetting). The grassland and wheat soils had more fungal-based and bacterial-based food webs, respectively (Supplementary Table S1). Field drought reduced bacterial-channel biomass (that is, bacteria and bacterial-feeding fauna) in both soils, but had contrasting effects on fungal-channel biomass (fungi and fungal-feeding fauna). In particular, field drought reduced fungal-energy-channel biomass 1 Soil and Ecosystem Ecology, Lancaster Environment Centre, Lancaster University, Lancaster LA1 3EX, UK, 2 University of Helsinki, Department of Environmental Sciences, Niemenkatu 73, FIN-15140, Lahti, Finland, 3 Biologisk Institut, Terrestrisk Økologi, Øster Farimagsgade 2D, 1353 København K, Denmark, 4 Southwest Biological Science Center, US Geological Survey, PO Box 5614, Building 56 No 150, Flagstaff, Arizona 86011, USA. *e-mail: [email protected]. NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange © 2012 Macmillan Publishers Limited. All rights reserved. 1 NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1368 LETTERS Resistance index a Fungal-channel biomass Wheat field Bacterial-channel biomass 1.00 PLFA evenness 0.98 0.96 0.94 0.92 0.90 0.88 Grassland Wheat field 0.86 Grassland Wheat field 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Microarthropod richness Grassland Wheat field Control Field drought Resilience index Grassland 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 Grassland Resilience index b 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 Wheat field 1 3 10 77 Time after rewetting (days) 1 3 10 77 Time after rewetting (days) 1 3 10 77 Time after rewetting (days) 1 3 10 77 Time after rewetting (days) Figure 1 | Resistance and resilience of soil food webs to a laboratory drought as affected by a previous field drought. a, Soil-food-web resistance to a laboratory-based drought as affected by a previous field drought. Field drought increased resistance of fungal-channel biomass and evenness of microbial PLFAs in grassland soil (field × drought interaction F1,12 = 9.1, P = 0.011, and F1,12 = 17.7, P = 0.001, respectively). Resistance of microarthropod richness was higher in grassland than in wheat-field soil (F1,12 = 10.9, P = 0.006). Resistance of fungal-channel biomass to drought was greater than resistance of bacterial-channel biomass (paired-samples T-test, t15 = 4.51, P < 0.001). b, Soil-food-web resilience following a laboratory-based drought as affected by a previous field drought. Field drought reduced resilience of the soil food web in grassland, but not in wheat-field soil (field × drought interactions for fungal-channel biomass F1,48 = 73.9, P < 0.001, bacterial-channel biomass F1,48 = 26.6, P < 0.001, evenness of microbial PLFAs F1,48 = 39.7, P < 0.001). Resilience of microarthropod richness was higher in wheat-field soil than in grassland soil (F1,48 = 29.4, P < 0.001). Resilience of fungal-channel biomass was lower than resilience of bacterial-channel biomass (paired-sample T-test, t63 = −6.42, P < 0.001). Markers denote treatment means ± 1 s.e.m. (n = 4). under wheat, but increased this measure, and the diversity and richness of microarthropods, in grassland (Supplementary Table S1). The response of microarthropods to field drought might have been caused by increased dissolved organic C and fungal biomass (Supplementary Table S1), as changes in resource availability are known to be key drivers of microarthropod community structure in grassland23 . In the laboratory, the fungal energy channel was more resistant (Fig. 1a), but less resilient (Fig. 1b), to drought than the bacterial energy channel, regardless of from which land use the soil originated. Also, after rewetting, bacterial, but not fungal, PLFAs decreased, indicating that bacteria are less resistant to rewetting than fungi (Supplementary Fig. S4). In addition, microarthropod richness had a greater resistance to the laboratory-based drought in grassland than wheat soil (Fig. 1a). Moreover, drought in the field significantly increased the resistance of the fungal energy channel and the evenness of microbial PLFAs to the second laboratory-based drought in grassland soil, but not in wheat soil. This increase in resistance after field drought suggests adaptation of the fungal-based food web of grassland soil to drought, but it was traded off by a decreased resilience to the laboratory-based drought (Fig. 1b). In general, our results indicate a trade-off between resistance and resilience in soil food webs, which is consistent with theoretical publications on resistance and resilience11,18,19 . 2 Losses of C (CO2 production) and N (leaching in drainage waters) were also more resistant to laboratory-based drought in grassland than in wheat soil (Fig. 2a). Grassland soil had greater amounts of readily available C and a higher absolute respiration in the field and at the start of the incubation experiment (Supplementary Fig. S2 and Table S1), and resistance of CO2 production to laboratory-based drought was higher in grassland than in wheat soil. The positive relationship between respiration and microbial growth (Supplementary Fig. S3), and the rapid increase of fungal and bacterial PLFAs after rewetting in grassland soil (Supplementary Fig. S4), indicate that the grassland microbial community more efficiently incorporated the flush of readily available C caused by drought15 . Adding to this, resilience of CO2 production was highest in the grassland field-control soil (Fig. 2b). Not only was N leaching from grassland soil more resistant to drought, but it was also lower during the whole experiment (including field-based measurements) than from wheat soil (Supplementary Fig. S5 and Table S1). This corresponds with the notion that N leaching is less from fungal-based than bacterial-based soils5,14 . Field drought decreased the resistance of N leaching to laboratory drought; this was traded off by an increased resilience (Fig. 2b). We were unable to calculate a resistance index of N2 O production owing to negative values in the control, which indicates N2 O consumption in some treatments24 . However, NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange © 2012 Macmillan Publishers Limited. All rights reserved. NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1368 Wheat field 3 10 77 Time after rewetting (days) 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 N leaching Control Field drought Grassland Wheat field 3 10 77 Time after rewetting (days) Resilience index Resistance index Grassland 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 Resilience index 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 CO2 Resilience index Resilience index 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 Resilience index b 0 ¬0.1 ¬0.2 ¬0.3 ¬0.4 ¬0.5 ¬0.6 ¬0.7 ¬0.8 ¬0.9 ¬1.0 Resilience index Resistance index a LETTERS 1.0 0.8 0.6 0.4 0.2 0 ¬0.2 ¬0.4 ¬0.6 ¬0.8 ¬1.0 N2O Grassland Wheat field 3 10 77 Time after rewetting (days) Figure 2 | Resistance and resilience of C and N losses to a laboratory drought as affected by a previous field drought. a, Resistance of CO2 production and N leaching to a laboratory-based drought as affected by a previous field drought. Resistance of both respiration and N leaching was higher in grassland than in wheat-field soil (F1,12 = 9.1, P = 0.011, and F1,12 = 33.7, P < 0.001, respectively). Resistance of N leaching was reduced by field drought (F1,12 = 26.7, P < 0.001), but more so in grassland soil (field × field-drought interaction F1,12 = 4.0, P = 0.069). b, Resilience of CO2 production, N leaching and N2 O production following a laboratory-based drought as affected by a previous field drought. Resilience of respiration at the end of the experiment was highest in grassland soil control (field × field-drought × sampling-date interaction F2,36 = 6.4, P = 0.004). Field drought increased resilience of N leaching (field × field-drought interaction F1,36 = 6.3, P = 0.017), but reduced resilience of N2 O production in grassland soil (field × field-drought interaction F1,36 = 6.3 and F1,36 = 4.2, P = 0.049, respectively). Markers denote treatment means ±1 s.e.m. (n = 4). laboratory drought significantly increased N2 O production in grassland (Supplementary Fig. S6), but not in wheat soil. This, and the higher N2 O production rates in grassland soil throughout the experiment, was probably a consequence of higher C availability. However, differences in soil structure and availability of anaerobic microsites, which is probably greater in grassland soil, might also have contributed25 . Still, rates of N2 O production recovered faster in grassland field-control soil than in wheat field-control soil following rewetting (Fig. 2b), which further points to more efficient incorporation of excess C in the fungal-based soil food web of grassland. We constructed a structural equation model to test the hypothesis that soil-food-web characteristics exerted a direct influence over soil C and N losses in the laboratory experiment (Supplementary Fig. S7). Because we wanted to focus on the interactions of soil biota and their relative influence on C and N losses across all treatments, we used residuals of the full-factorial analysis of variance model for the modelling (Supplementary Methods). Because of this step, variation in the data caused by treatment effects or biophysical differences among wheat fields and grasslands (for example soil nutrient and C availability) cannot influence the outcome of our model. After these treatment effects had been removed, the residual variance in soil-food-web characteristics and C and N fluxes (ranging from 10.9 to 64.7%, Fig. 3) fed into the structural equation model. A significant proportion of this variance in C and N fluxes was explained by variance in soil-food-web characteristics (Fig. 3). Throughout the laboratory experiment, a high fungal/bacterial channel ratio was associated with lower amounts of C lost through respiration (Fig. 3a–d), which is consistent with our observation that the more fungal-based grassland microbial community incorporated C more efficiently. Greater evenness of microbial PLFAs was associated with reduced loss of N through leaching, and after 10 days (Fig. 3c) increased microarthropod richness. Increased microarthropod richness can stimulate N mineralization26 , which here—in the absence of plant N uptake—led to increased N leaching from soil (Fig. 3d). In addition, respiration—a measure of heterotrophic microbial activity—strongly controlled N cycling. Respiration positively influenced N2 O production at the first sampling date, suggesting that C availability controlled denitrification. It also exerted a negative control on N leaching across all sampling dates (Fig. 3). In conclusion, our data show that differences in soil food webs resulting from agricultural land use have significant implications for their resistance and resilience under drought, and that this has consequences for the loss of C and N from soil. Specifically, we show NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange © 2012 Macmillan Publishers Limited. All rights reserved. 3 NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1368 5 F/B channel ratio ¬0.29P < 0.001 ¬0.08 0.0 5 .08 ¬0 ¬0 .0 3 R2 = 0.02 ¬0.14P = 0.04 Respiration PLFA evenness ¬0.10P = 0.03 5 2 0.0 0.0 8 0.0 R2 = 0.12 Total N leached ¬0.06 0.0 0 ¬0.39P < 0.001 0.08 R2 = 0.04 N2O production 0.17 P= 0 R2 = 0.01 Microarthropod richness R2 = 0.14 Total N leached ¬0.16P = 0.03 3 0.4 ¬0.16P = 0.03 5 0.0 PLFA evenness 2 .01 ¬0 0.1 2 R2 = 0.07 Respiration 1 0.0 0.1 Respiration ¬0.27P < 0.001 ¬0.25P = 0.04 77 days R2 = 0.05 ¬0.20P = 0.04 0 ¬0 .06 ¬0.05 0..0 5 0.08P = 0.006 .08 ¬0 .05 PLFA evenness d .08 ¬0 ¬0 ¬0.08 7 0.0 ¬0.13 0 0.08 R2 = 0.03 N2O production 0.17 R2 = 0.04 N2O production 0.17 5 0.0 R2 = 0.01 R2 = 0.01 Microarthropod richness F/B channel ratio R2 = 0.25 Total N leached ¬0.26P = 0.03 Microarthropod richness F/B channel ratio 4 0.0 .01 ¬0 0.1 1 3 days R2 = 0.02 Respiration PLFA evenness 10 days ¬0 .03 b 0.10 .06 ¬0 ¬0.14P = 0.04 F/B channel ratio c R2 = 0.32 N2O production ¬0.17 0.51P < 0.001 R2 = 0.02 Microarthropod richness ¬0.24P < 0.001 1 day 0.08 a ¬0.07 LETTERS R2 = 0.30 Total N leached Figure 3 | Relationships between remaining variance in soil-food-web characteristics and C and N fluxes, after variance accounted for by experimental treatments has been removed, 1, 3, 10 and 77 days after rewetting. Residual variances that fed into the structural equation model were as follows: microarthropod richness 42.2%, F/B channel ratio 52.2%, PLFA evenness 19.9%, respiration 21.3%, N2 O 64.7%, total N leached 10.9%. The weight of the arrows indicates the strength of the causal relationship, supplemented by a path coefficient. R2 values denote the amount of variance explained by the model for the response variables. Our overall model fit was satisfactory (χ 2 = 64.1, P = 0.24; RMSEA = 0.033, P = 0.754). that fungal-based food webs of grassland soil, and the processes of C and N loss that they govern, are more resistant to drought than are bacterial-based food webs of more intensively managed wheat soil. Importantly, the resistance of the grassland soil food web was increased after the field drought, which suggests that it adapts to a changing climate. Moreover, we show that, across both land-use systems, more fungal-based soil food webs and a greater evenness of microbial PLFAs mitigated the effects of drought on processes of C and N loss from soil. Collectively, our findings show that land use alters the stability of soil food webs and the ecosystem services that they deliver under climate change. were calculated as RS(t0 ) = 1 − (2|D0 |)/(C0 + |D0 |), with C0 the value of the control at the end of the disturbance and |D0 | the absolute difference between the control and the disturbed soil22 . Resilience indices were calculated as RL(tx ) = (2|D0 |)/(|D0 | + |Dx |) − 1, with |Dx | the absolute difference between the control and the disturbed soil at time x (ref. 22). Resistance indices were analysed using two-way analysis of variance with factors land use and field drought; resilience indices were analysed using three-way analysis of variance with factors land use, field drought and sampling time, using PASW Statistics 17.0 (2009 SPSS). We constructed our structural equation model in Amos 18.0 (2009 SPSS), using the multigroup modelling approach to track changes in the strength of the pathways in the model. We confirmed adequate model fit with the χ 2 test and the RMSEA index. A full description of the modelling procedure is available in Supplementary Methods. Methods Received 28 February 2011; accepted 6 December 2011; published online 29 January 2012 The field experiment was located in the south of England (51◦ 320 55.2000 N, 1◦ 040 37.4400 W) and consisted of grassland and wheat field on the same slope and the same soil type. Field drought was simulated by randomly placing three transparent roofs (1.6 m × 1.6 m) in each field for three months (April–July 2009). After this, soil was collected from droughted and control plots. Soil from each treatment was transferred into 1 l pots, rewetted to 60% water holding capacity (WHC), and used for the full-factorial drought experiment (field × field drought × laboratory drought × 5 sampling dates, each treatment replicated four times, resulting in 160 pots) in the greenhouse (20 ◦ C, day/night 16 h/8 h, randomized block design). During two weeks, drought pots were dried to, and kept at, 20% WHC versus control pots that were kept at 60% WHC. Successively, all pots were rewetted and kept at 60% WHC for 2.5 months. A full description of the set-up of the field and the laboratory experiment, N and C loss measurements, and soil microbial and faunal analyses and biomass calculations is available in the Supplementary Methods. Resistance indices 4 References 1. Godfray, H. C. J. et al. Food security: The challenge of feeding 9 billion people. Science 327, 812–818 (2010). 2. Gornall, J. et al. Implications of climate change for agricultural productivity in the early twenty-first century. Phil. Trans. R. Soc. B 365, 2973–2989 (2010). 3. Power, A.G. Ecosystem services and agriculture: Tradeoffs and synergies. Phil. Trans. R. Soc. B 365, 2959–2971 (2010). 4. Bardgett, R. D. & Cook, R. Functional aspects of soil animal diversity in agricultural grasslands. Appl. Soil Ecol. 10, 263–276 (1998). 5. De Vries, F. T., Hoffland, E., van Eekeren, N., Brussaard, L. & Bloem, J. Fungal/bacterial ratios in grasslands with contrasting nitrogen management. Soil Biol. Biochem. 38, 2092–2103 (2006). NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange © 2012 Macmillan Publishers Limited. All rights reserved. NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1368 6. Postma-Blaauw, M. B., de Goede, R. G. M., Bloem, J., Faber, J. H. & Brussaard, L. Soil biota community structure and abundance under agricultural intensification and extensification. Ecology 91, 460–473 (2010). 7. IPCC Climate Change 2007: The Physical Science Basis (eds Solomon, S., et al.) (Cambridge Univ. Press, 2007). 8. Bardgett, R. D., Freeman, C. & Ostle, N. J. Microbial contributions to climate change through carbon cycle feedbacks. ISME J. 2, 805–814 (2008). 9. Bardgett, R. D. The Biology of Soil: A Community and Ecosystem Approach (Oxford Univ. Press, 2005). 10. Bardgett, R. D. & McAlister, E. The measurement of soil fungal:bacterial biomass ratios as an indicator of ecosystem self-regulation in temperate meadow grasslands. Biol. Fertil. Soils 29, 282–290 (1999). 11. Hedlund, K. et al. Trophic interactions in changing landscapes: Responses of soil food webs. Basic Appl. Ecol. 5, 495–503 (2004). 12. Wardle, D. A. Controls of temporal variability of the soil microbial biomass: A global-scale synthesis. Soil Biol. Biochem. 30, 1627–1637 (1998). 13. De Vries, F. T., Van Groenigen, J. W., Hoffland, E. & Bloem, J. Nitrogen losses from two grassland soils with different fungal biomass. Soil Biol. Biochem. 43, 997–1005 (2011). 14. Gordon, H., Haygarth, P. M. & Bardgett, R. D. Drying and rewetting effects on soil microbial community composition and nutrient leaching. Soil Biol. Biochem. 40, 302–311 (2008). 15. Six, J., Frey, S. D., Thiet, R. K. & Batten, K. M. Bacterial and fungal contributions to carbon sequestration in agroecosystems. Soil Sci. Soc. Am. J. 70, 555–569 (2006). 16. Wilson, G. W. T., Rice, C. W., Rillig, M. C., Springer, A. & Hartnett, D. C. Soil aggregation and carbon sequestration are tightly correlated with the abundance of arbuscular mycorrhizal fungi: Results from long-term field experiments. Ecol. Lett. 12, 452–461 (2009). 17. Vauramo, S. & Setälä, H. Urban belowground food-web responses to plant community manipulation—impacts on nutrient dynamics. Landsc. Urban Plan. 97, 1–10 (2010). 18. Pimm, S. L. The complexity and stability of ecosystems. Nature 307, 321–326 (1984). 19. Grime, J. P. Plant Strategies, Vegetation Processes, and Ecosystem Properties (Wiley, 2001). 20. Orwin, K. H., Wardle, D. A. & Greenfield, L. G. Context-dependent changes in the resistance and resilience of soil microbes to an experimental disturbance for three primary plant chronosequences. Oikos 112, 196–208 (2006). LETTERS 21. Holtkamp, R. et al. Soil food web structure during ecosystem development after land abandonment. Appl. Soil Ecol. 39, 23–34 (2008). 22. Orwin, K. H. & Wardle, D. A. New indices for quantifying the resistance and resilience of soil biota to exogenous disturbances. Soil Biol. Biochem. 36, 1907–1912 (2004). 23. Cole, L., Buckland, S. M. & Bardgett, R. D. Relating microarthropod community structure and diversity to soil fertility manipulations in temperate grassland. Soil Biol. Biochem. 37, 1707–1717 (2005). 24. Chapuis-Lardy, L., Wrage, N., Metay, A., Chotte, J. L. & Bernoux, M. Soils, a sink for N2 O? A review. Glob. Change Biol. 13, 1–17 (2007). 25. Cavigelli, M. A. & Robertson, G. P. The functional significance of denitrifier community composition in a terrestrial ecosystem. Ecology 81, 1402–1414 (2000). 26. Liiri, M., Setälä, H., Haimi, J., Pennanen, T. & Fritze, H. Relationship between soil microarthropod species diversity and plant growth does not change when the system is disturbed. Oikos 96, 137–149 (2002). Acknowledgements This project was part of the EU Seventh Framework funded SOILSERVICE project, led by K. Hedlund. We thank all project partners for contributing to this manuscript through discussions. We thank S. Mortimer and D. Carpenter for setting up the field experiment, and G. Hildred for allowing us into his fields. H. Quirk, L. Trimnell, V. van Velzen, A. Spangenberg, L. F. Petersen, I. Dodd, G. Mies, F. Willeboordse, B. v/d Waterbeemd, C. Siderius, E. Wilson and K. Wilson helped with field and laboratory work. We thank K. Orwin and W. van der Putten for commenting on the manuscript. Author contributions R.D.B., H.M.S., S.C., F.T.d.V., M.E.L. and L.B. had the original idea for the experiment. F.T.d.V. set up the experiment, and laboratory work was conducted by F.T.d.V., M.E.L. and L.B. M.A.B. carried out the structural equation modelling. The manuscript was written principally by F.T.d.V. and R.D.B., with extensive input from H.M.S., S.C., M.E.L., L.B. and M.A.B. Additional information The authors declare no competing financial interests. Supplementary information accompanies this paper on www.nature.com/natureclimatechange. Reprints and permissions information is available online at http://www.nature.com/reprints. Correspondence and requests for materials should be addressed to F.T.d.V. NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange © 2012 Macmillan Publishers Limited. All rights reserved. 5
© Copyright 2026 Paperzz