A decade of free-air CO2 enrichment increased the carbon throughput in a grass-clover ecosystem but did not drastically change carbon allocation patterns

Authors


Summary

  1. The response of the soil carbon cycle to increasing atmospheric CO2 concentration has far reaching consequences for the ecosystem carbon balance under future climatic conditions. We report on work carried out in the Swiss free-air CO2 enrichment (FACE) experiment, where we used in situ 13CO2 labelling to determine whether elevated CO2 (+230 μL L−1) concentration changes the fate of recently assimilated carbon in the soil microbial community.

  2. Elevated CO2 (eCO2) concentration had an overall positive effect on microbial abundance (< 0·001) with the gram-negative bacteria showing significantly increased quantities.

  3. Gram-negative bacteria and saprotrophic fungi tended to utilize a higher amount of recently assimilated carbon under eCO2. Arbuscular mycorrhizal fungi (AMF) utilized plant-assimilated carbon within 1 day after the 13CO2 pulse and 13C uptake patterns in AMF suggest that carbon transfer is faster under eCO2 concentration than under ambient CO2 (aCO2). Additionally, the respiration of recently assimilated carbon was significantly higher under eCO2 than aCO2 concentration.

  4. Our data suggest that elevated atmospheric CO2 concentration accelerated and increased the utilization of recently assimilated carbon by the microbial community without changing the microbial community composition drastically.

  5. We conclude that a higher standing soil microbial biomass under eCO2 concentration was the key cause for the higher carbon flow through the plant–soil system. Carbon utilization by microbial functional groups was only little affected by a decade of CO2 enrichment.

Introduction

Human-induced environmental change is a major societal concern (IPCC 2007). Predicting and mitigating the impacts of human activity on the Earth's climate is therefore receiving considerable attention. In particular, understanding how terrestrial ecosystems will respond to elevated atmospheric CO2 (eCO2) concentration is a priority research area in view of the existing feedbacks between the global carbon cycle and climate change (Luo et al. 1996; Carney et al. 2007; Selsted et al. 2012).

The impact of eCO2 concentration on carbon cycling has been investigated on small-scale mesocosms (Sowerby et al. 2000; Carney et al. 2007; Drigo et al. 2013), but also in large-scale field experiments (Van Kessel et al. 2000a; Denef et al. 2007; Hill et al. 2007; Selsted et al. 2012). One of the first free-air CO2 enrichment (FACE) experiments was the Swiss FACE that aimed to understand the impact of eCO2 concentration in combination with nitrogen (N) fertilization and management practice on grass-clover vegetation using Lolium perenne L. and Trifolium repens L. (Hebeisen et al. 1997). It has been reported that after more than 8 years, photosynthesis was increased in both species (Ainsworth et al. 2003a,b), which is in accordance with responses in other grasses (Albert et al. 2011) and forbs (Ainsworth & Long 2005). Increased uptake of atmospheric CO2 has frequently led to the hypothesis that the increased carbon flow into the ecosystem could increase soil carbon sinks, but results are ambiguous.

Extensive research in the Swiss FACE showed that increased photosynthesis did not necessarily lead to a higher carbon storage potential of the soil (Sowerby et al. 2000; Van Kessel et al. 2000a; Drissner et al. 2007) and this has also been reported by others (Fitter et al. 1996; Carney et al. 2007; Selsted et al. 2012). This result is not surprising because increased carbon availability in the ecosystem does not only affect the carbon accumulated in plant biomass (shoots and roots; Hebeisen et al. 1997; Suter et al. 2002) but also assimilated carbon released into the soil matrix, which stimulates microbial activity and maintains plant available nutrient levels in soil (Fontaine, Mariotti & Abbadie 2003; Paterson et al. 2007; Talbot, Allison & Treseder 2008). A stimulation of the microbial biomass results in carbon being bound in the microbial community, and at the same time, these microbes mineralize soil organic matter (SOM) and release carbon by respiratory processes (Carney et al. 2007).

In the Swiss FACE, the microbial community has been intensively studied (Hartwig et al. 2000; Lüscher et al. 2000; Staddon, Jakobsen & Blum 2004; Drissner et al. 2007). The microbial biomass was stimulated under eCO2 concentration (Sowerby et al. 2000; Drissner et al. 2007), but increased microbial activity led to an enhanced soil CO2 efflux counterbalancing the carbon input (Sowerby et al. 2000). Increased soil CO2 efflux has also been observed for a temperate heathland where eCO2 in combination with warming and periodic summer drought induced soil carbon loss by 21% (Selsted et al. 2012).

One major microbial group associated with clover is the nitrogen-fixing gram-negative rhizobia bacteria that were found to balance soil nutrient levels under eCO2 concentration in the Swiss FACE (Lüscher et al. 2000; Hartwig et al. 2002). In the grass-clover mixture, 59% of the nitrogen was supplied from atmospheric nitrogen fixation under ambient CO2 (aCO2) concentration, and this amount was increased by 12% when plants were exposed to 600 ppm CO2 (Zanetti et al. 1996).

Fungi were another important microbial group in the Swiss FACE: fungal enzyme production was increased under eCO2 concentration and fungal abundance rose under low nitrogen fertilization (Drissner et al. 2007). Fungi have been shown to be a relevant driver of soil CO2 efflux under eCO2 concentrations (Carney et al. 2007) due to their ability to increase nutrient availability in the soil by mineralizing SOM (Carney et al. 2007; Drissner et al. 2007). Another important microbial group in terms of nutrient balance of an ecosystem is the group of mycorrhizal fungi. Mycorrhizal fungi provide nutrients, especially phosphorus, to plants in exchange for plant carbon. This is especially important in the rhizosphere where phosphorous can be depleted and mycorrhizal hyphae can reach beyond the zone of nutrient depletion. In the Swiss FACE, arbuscular mycorrhizal fungi (AMF) were positively affected by eCO2 conditions combined with low nitrogen fertilization (Staddon, Jakobsen & Blum 2004).

Here, we used an in situ 13CO2 pulse labelling approach (Ostle et al. 2000; Leake et al. 2006) to track and quantify the carbon flow through the ecosystem under ambient and elevated atmospheric CO2 concentrations. We measured the 13C enrichment of microbial fatty acid biomarkers to determine the effect of eCO2 on the relative activity of major soil microbial functional groups. In particular, the 13C label content of phospholipid fatty acids (PLFA) and neutral lipid fatty acids (NLFA) can be used to compare the relative activity of different groups of soil microorganisms, for example, gram-negative and gram-positive bacteria, saprotrophic and AMF (Treonis et al. 2004; Olsson & Johnson 2005). Elevated CO2 is known to affect soil microbial communities, principally via altered carbon input quality, and clearly affects the relative activity of different soil micro-organisms (Montealegre et al. 2002). We hypothesize that elevated CO2 concentration changes the fate of carbon in the soil microbial community towards a higher carbon recovery in gram-negative bacteria, saprotrophic fungi and AMF.

Materials and methods

The Swiss Face Experiment

The research was carried out at the Swiss FACE site, Lindau, Switzerland (Hebeisen et al. 1997), where elevated atmospheric CO2 concentration (+230 μL L−1) was imposed for 10 years from 1993. The lowland grassland site and treatments are fully described elsewhere (Schortemeyer et al. 1996). There were three replicate FACE rings and three replicate control (ambient) rings yielding a total of six labelling chambers. The labelling work reported here was entirely carried out in the mixed L. perenne L. and T. repens plots under low N addition (14 g m−2 year−1).

Pulse Labelling with 13CO2

The 13C label was delivered to the vegetation as 13CO2 at 99 atom% 13C (Staddon et al. 2003) at two CO2 concentrations, that is, ambient (aCO2, 370 μL L−1) and elevated (eCO2, 600 μL L−1). This was achieved using an inexpensive and highly flexible method developed for labelling in the field, which, importantly, allowed the delivery of the 13C label simultaneously at different CO2 concentrations. Briefly, this method involved six individual labelling units consisting of 250 L capacity gas impermeable Tedlar bags, which were filled with air containing 13CO2 at 370 or 600 μL L−1. The 99 atom% 13CO2 was mixed into bottled CO2-free air to achieve the required concentration. The labelled air was then pumped at a flowmeter controlled rate of 1·0 L min−1 to purpose built 20 cm diameter, 15 cm high, transparent acrylic labelling chambers, placed flush with the soil surface. The labelling areas were selected randomly within the low N mixed vegetation plots. Exhaust air from the labelling chambers was expelled 10 m outside the FACE and control rings. Labelling was carried out on 13 September 2002 in the afternoon and lasted for 3·5 h under sunny conditions.

We tested whether 13C from the 13CO2 labelling pulse entered the soil directly in significant quantity by applying a 13CO2 pulse to vegetation that was kept dark (labelling chambers covered in opaque plastic; Staddon et al. 2003). Labelling in darkened chambers was carried out outside the FACE rings. Gas samples in unlabeled area were taken at least 2 m away from the labelled area. Soil-respired CO2 was collected in the same manner as for the actual labelling experiment (detailed below). We found that the 13C content of soil-respired CO2 in unlabeled area and the labelled darkened area was similar at the relevant times after labelling.

Gas, Plant and Soil Sample Collection

Gas sampling was carried out to investigate the 13C content in soil CO2 efflux 2 h and 1, 2 and 4 days after 13CO2 labelling. Samples were taken within the labelled area and at 5 and 15 cm away from the outer edge of the labelled area. Soil-respired CO2 was collected using one 10 mL Exetainer vial per sampling point (Labco Ltd., High Wycombe, UK) that was placed upside down on the soil surface (Staddon et al. 2003). Vials were flushed with ambient air before the incubation so that, within treatments, they all had the same starting point. Care was taken not to include shoot respiration within the trapped soil-respired CO2. The tubes were left in place for 4 h, after which, each tube was carefully lifted from the soil surface and capped. The soil-respired CO2 samples were stored at ambient temperature.

Samples of above-ground biomass were taken in all labelled areas to investigate the 13C content 2 h, 1 and 2 days after 13CO2 labelling. Samples were taken within the labelled area and were obtained as close as possible to where the Exetainer vials were located.

Soil cores (2 cm diameter) were taken in all labelled areas to examine the fate of recently assimilated carbon (13C) into the soil microbial community and to assess the microbial biomass 2 h and 1, 2, 3, 4, 6 and 14 days after the 13CO2 pulse. Cores were taken to a depth of 10 cm at the same location as the gas samples. Above-ground biomass samples and soil cores were frozen at −18 °C until analysis.

Carbon-13 Analyses of Gas and Plant Samples

The 13C content of CO2 (soil respiration) was analysed in subsamples transferred to evacuated 2 mL crimp seal vials for immediate processing. The vial headspace was purged into the He carrier stream of a PreCon trace gas preparation-concentration unit (Thermo Electron, Bremen, Germany) interfaced with a HP 6890 gas chromatograph coupled in continuous flow mode to a Finnigan MAT Delta PLUS isotope ratio mass spectrometer (Thermo Electron). As working standard, we used commercial CO2 calibrated against certified 13CO2 standards (Messer Griesheim, Krefeld, Germany).

Total carbon and the 13C/12C ratio in plant material was measured by Dumas combustion on an elemental analyser (CE 1110; Thermo Electron, Milan, Italy) coupled in continuous flow mode to the Delta PLUS. Analyses were performed on dried, finely ground 2–3 mg samples weighed out into tin combustion capsules. As working standard for 13C determinations, commercial CO2 calibrated against certified 13C-sucrose material (IAEA, Vienna, Austria) was used.

Extraction, Identification and 13C Analyses of Fatty Acids

The extracted PLFAs were selected as being specific to various groups of soil micro-organisms: general biota [16:0, 18:0 (both PLFAs synthesized by microbes and plants)], gram-negative bacteria (cy17:0, cy19:0, 18:1ω7), gram-positive bacteria (i15:0, a15:0, i16:0, i17:0, a17:0, 10Me17), actinomycetes (10Me16, 10Me18) and saprotrophic fungi (18:1ω9, 18:2ω6,9). The NLFA 16:1ω5 as specific to AMF was chosen as it has been shown to be a more sensitive indicator of AMF hyphae in soil than the PLFA 16:1ω5 (Olsson 1999).

Lipids within soil (roots removed) were extracted in a one-phase mixture of citrate buffer, methanol and chloroform (0·8:2:1, v/v/v, pH 4·0). The lipids were fractionated into neutral lipids, glycolipids and phospholipids on pre-packed silica columns (100 mg sorbent mass, Varian Medical Systems, Palo Alto, CA, USA) as described by van Aarle & Olsson (2003). The fatty acid residues in neutral lipids and phospholipids were transformed into free fatty acid methyl esters and identified and quantified by gas chromatography.

The 13C enrichment in fatty acid methyl esters was determined in a 20-20 IRMS (PDZ Europa Scientific Instruments, Crewe, UK) interfaced with a Hewlett Packard 6890 gas chromatograph (Hewlett Packard, Palo Alto, CA, USA). The chromatographic conditions were as described by Olsson et al. (2005). The 13C values were calculated based on atom% 13C of the reference CO2 gas, injected three times at the beginning and end of a chromatographic run. The reference CO2 was standardized with the Pee Dee Belemnite (PDB) standard. The precision of the reference gas 13C was 0·2‰. Integration for each peak was checked and corrected manually.

Carbon isotopic enrichment was highest in above-ground vegetation and was expressed as atom% 13C excess (APE) and calculated as follows:

display math( Eqn 1)

where the 13C contentsample is the measured 13C content in above-ground vegetation after the 13CO2 pulse, and 13C contentcontrol is the 13C content in above-ground vegetation before the experiment. Measured 13C enrichments (Δ13C) in fatty acids and soil respiration are calculated as:

display math( Eqn 2)

where δ13Csample is the 13C/12C ratio of individual PLFAs and soil respiration after the 13CO2 pulse and δ13Ccontrol before the 13CO2 pulse.

Data Analyses

Statistical analyses were performed using R version 2.12.1 (www.R-project.org). All data were checked for normality and where necessary log-transformed. Analysis of variance (anova) was performed to test for effects of CO2 concentration on 13C content in above-ground plant biomass. Sampling time was included as fixed variable to test for the effect of time.

Fatty acid quantities are presented as averages of sampling times and replicates (FACE rings) because fatty acid abundances are not assumed to change over the investigated time period. Standard errors were calculated using error propagation. The overall effect of eCO2 on fatty acid abundances was tested using a manova, and the effect of eCO2 on individual fatty acids was analysed with a linear mixed-effect model (lme) using ‘FACE ring’ as a random variable because fatty acid quantities were consistently lower in one of the FACE rings.

The effect of eCO2 on Δ13C in microbial functional groups (averages of 13C values of functional group specific fatty acids) was analysed using the ‘lme’ function with ‘FACE ring/time’ as random variable. The model was extended for a ‘varIdent’ structure that accounts for inhomogeneous variances if the data did not show homogeneous variances. The same model was used to test for the effect of CO2 concentration on soil-respired Δ13CO2.

Results

Plant 13C content decreased with time but plant 13C enrichment was not different between CO2 treatments (Fig. 1). Microbial fatty acid abundances were significantly higher under eCO2 compared to aCO2 concentration (anova:< 0·001). This CO2-induced increase was lowest for actinomycetes (9 ± 4%) and highest for AMF (40%; Fig. 2). Gram-negative-specific cy17:0 and cy19:0 PLFAs were significantly more abundant in the eCO2 treatment (< 0·05). The same trend was observed for the gram-positive-specific biomarkers i15:0 (= 0·098) and i16:0 (P = 0·065) and the actinomycetes PLFA 10Me16:0 (= 0·057). The AMF-specific NLFA 16:1ω5 also tended to be more abundant under eCO2 (= 0·061).

Figure 1.

Above-ground 13C enrichment at ambient and elevated CO2 concentration over time. 13C enrichment is expressed as 13C atom% excess. Effect of the CO2 treatment was statistically analysed with an analysis of variance (anova) and is given in the upper right corner. Enrichments are presented as means ± SE.

Figure 2.

Quantities of microbial phospholipid fatty acid (nmol g−1 dry soil) and the neutral lipid fatty acid 16:1ω5 under ambient and elevated CO2 concentrations. The effect of the eCO2 treatment is given in the upper right corner. Bars represent means ± SE. *indicates significant differences on a level of P ≤ 0·05.

The Δ13C in fatty acids was affected by functional group (= 0·046) and CO2 concentration (P = 0·025) when considering all time points. There was no significant effect of eCO2 on Δ13C in microbial functional groups. However, eCO2 had a tendency to increase the Δ13C in gram-negative bacteria and saprotrophic fungi, whereas AMF were more enriched in 13C when exposed to aCO2 concentration (Fig. 3). The Δ13C in gram-positive bacteria (= 0·40) and actinomycetes (= 0·50) were unaffected by eCO2 (Fig. S1, Supporting information).

Figure 3.

Δ13C (‰) in microbial functional groups at ambient and elevated CO2 concentrations over time: (a) gram-negative bacteria (average of cy17:0, cy19:0 and 18:1ω7), (b) saprophytic fungi (average of 18:1ω9 and 18:2ω9) and (c) arbuscular mycorrhizal fungi neutral lipid fatty acids (NLFA 16:1ω5). Effects of the CO2 treatment are given in the upper right corners. Data are presented as means ± SE.

There was a substantial difference in Δ13C of soil-respired CO2 between aCO2 and eCO2 within the labelling plots (Fig. 4). The 13C enrichment of soil-respired CO2 was considerably higher at eCO2 than aCO2. At 5 cm distance from the labelled plots, the 13CO2 efflux pattern was similar at aCO2 and eCO2 over the 4-day sampling period (data not shown). Δ13C was on average 4‰ in aCO2 and was increased by ~2·5‰ under eCO2 concentration. At 15 cm distance from the labelled plots, aCO2 and eCO2 plots showed a ~5‰ Δ13C in soil CO2 efflux. In aCO2 plots, the measured Δ13C increased to ~8‰ and approached zero thereafter. In eCO2 plots, measured Δ13C approached zero already 1 day after the labelling pulse.

Figure 4.

Soil-respired Δ13C (‰) at ambient and elevated CO2 concentration. Effect of the CO2 treatment is given in the upper right corner. Data are represented as means ± SE.

Discussion

Plant Carbon Uptake

Our results show the same 13C enrichment of above-ground biomass under aCO2 and eCO2 concentrations (Fig. 1). However, we can assume that the uptake of 13CO2 was higher under eCO2 conditions because we know from earlier studies at the Swiss FACE that eCO2 concentration affects the photosynthetic capacity of plants and results in increased carbon uptake (Ainsworth et al. 2003a,b). Seen in this context, our results suggest that photo-assimilated carbon is not stored in above-ground biomass but transported below-ground under the assumption that plant biomass is similar or higher under eCO2 compared to aCO2 (Ainsworth & Long 2005; Kongstad et al. 2012). This is a reasonable assumption because the labelling pulse was deployed in late summer when temperate plants increase the allocation of carbon to below-ground parts to prepare for the winter season. Increased fixation of CO2 and a more rapid transfer of photo-assimilates below-ground under eCO2 concentration (Hebeisen et al. 1997; Daepp, Nösberger & Lüscher 2001) create a bigger carbon pool for plant and below-ground activity but, at the same time, increase the nutrient demand for plants and microbes.

Microbial Community Response to Elevated CO2

Elevated CO2 significantly increased the soil microbial PLFA abundance by ~16% compared to aCO2 concentration (Fig. 2). This finding is in accordance with previous results from the same experimental site where total PLFA abundances were increased by about 25% (Drissner et al. 2007). Microbial PLFA abundance and thus biomass (Frostegård & Bååth 1996) likely increased due to the higher carbon input to the soil that stimulated microbial growth and activity due to, for example, rhizodeposition.

In particular, the abundance of gram-negative bacteria was significantly increased under eCO2 concentration. Gram-negative bacteria are fast growing and highly dependent on plant photo-assimilates (Paterson et al. 2007; Jin & Evans 2010; Garcia-Pausas & Paterson 2011) and the increase of root growth (Jongen et al. 1995; Arndal et al. 2013) and connected rhizodeposition under eCO2 can therefore have a facilitating effect on gram-negative bacteria.

In soil, the free-living gram-negative bacteria Azotobacter spp. (Strandberg & Wilson 1968) and the symbiotic (gram-negative) rhizobia are able to fix atmospheric nitrogen, and thus, they are a key component in the nitrogen balance of the soil. However, in the present study, it was not possible to determine whether these specific groups were part of the general increase in gram-negative bacteria biomass. In general, in other studies from the Swiss FACE, nodule formation was increased by 17–50% due to the symbiosis between plants and rhizobia under eCO2 concentration (Schortemeyer et al. 1996; Montealegre et al. 2000). Nitrogen-fixing bacteria counterbalanced the expected negative effect of decreased above-ground biomass and lower nitrogen yield and maintained a similar C/N balance under aCO2 and eCO2 concentrations (Hartwig et al. 2000; Lüscher et al. 2000). Additionally, the increased nitrogen demand arising from increased carbon input under eCO2 concentration was entirely supplied by nitrogen-fixing symbionts in the grass-clover mixture (Zanetti et al. 1996). However, the utilization of recently assimilated carbon by free-living nitrogen-fixing bacteria should be investigated in greater detail to confirm their dependence on rhizodeposits.

The AMF biomarker (NLFA 16:1ω5) tended to be positively affected by eCO2 concentration. Even though this result was not significant, this trend is not a random effect because the extraradical mycorrhizal hyphae density was increased from 0·3 to 0·8 m g−1 soil in aCO2 to 1·0–1·7 m g−1 soil in eCO2 by October 2002, (Staddon, Jakobsen & Blum 2004) that is, 1 month after the pulse labelling. AMF are known to act as a carbon sink (Jakobsen & Rosendahl 1990; Olsson & Johnson 2005), and due to their symbiotic connection with the host plant, AMF are the first group to receive recently plant-assimilated carbon (Olsson & Johnson 2005).

Changed Microbial Activity

Arbuscular mycorrhizal fungi displayed distinct 13C enrichment in the biomarker NLFA 16:1ω5 under labelling at the two CO2 concentrations (Fig. 3c). Under aCO2 concentration, uptake of recently assimilated CO2 peaked at 80‰ δ13C between 2 and 3 days after labelling and decreased thereafter. In contrast, under eCO2 concentration, the maximum δ13C was measured to be ~44‰ 1 day after labelling. However, the lower Δ13C in the eCO2 treatment is a result of pool dilution because extraradial mycorrhizal hyphae density was ~50% increased (Staddon, Jakobsen & Blum 2004) in soils with eCO2 concentration. The measured isotopic value must therefore be diluted by 50% when rhizodeposition was similar in eCO2 and aCO2 treatments. When considering the isotopic dilution effect, the level of NLFA would be similar.

However, carbon transport has been shown to be faster under eCO2 than under aCO2 concentration (Selsted et al. 2012). Furthermore, the symbiosis between AMF and plant roots facilitate a very fast transfer of recently assimilated carbon between plants and AMF (Moyano, Kutsch & Schulze 2007) in exchange for nutrients (Jakobsen & Rosendahl 1990; Johnson, Leake & Read 2002; Olsson & Johnson 2005). In a 14C labelling experiment, 14C-labelled carbon that was taken up by plants was re-allocated below-ground and peaked in associated AMF within 16 h after the labelling pulse (Jakobsen & Rosendahl 1990). Our second sampling took place 1 day after the labelling implying that our measuring window was too wide to detect the early peak of enrichment in the eCO2 treatment under fast carbon transfer conditions. Under aCO2 concentration, carbon transfer is slower than under eCO2 because the labelling peak seems to occur between 1 and 2 days after the labelling pulse. Thus, the symbiotic lifestyle of AMF results in a rapid transfer of carbon from the plant, and this carbon transfer is faster under eCO2 than under aCO2 concentrations. A more detailed evaluation of the impact of AMF on the net carbon balance of the system could be achieved with more frequent samplings.

Fungi did not take up much of the recent plant-assimilated carbon under aCO2 concentration (Fig. 3b). In contrast, fungi grown under eCO2 concentration showed a dynamic 13C uptake. At the Swiss FACE experiment, it has been reported that fungal and gram-negative-specific PLFA abundance were reduced under high nitrogen fertilization compared to the low fertilization treatment (Drissner et al. 2007). Thus, under low nitrogen fertilization, the abundance of fungi and gram-negative bacteria is important for, for example, maintaining the nutrient balance under induced nutrient stress in the eCO2 treatment. In addition, soil enzyme activities were increased by 17–42% under eCO2 concentration in October 2002 (Drissner et al. 2007), suggesting a crucial role of the fungal group in the maintenance of available nutrients.

Gram-negative bacteria under eCO2 concentration received recently assimilated 13C after a time delay of about 3 days (Fig. 3a). The observed peak enrichment at day three fits to the observation that carbon is first utilized by AMF and becomes available from rhizodeposits for bacteria and other soil organisms later (Olsson & Johnson 2005). The significant increase in gram-negative biomarker quantities coupled with a higher Δ13C indicates a high carbon demand in that microbial group which could be related to the effect of eCO2 on rhizodeposition and the facilitation of gram-negative bacteria or to nitrogen-fixing bacteria that have been shown to be abundant in this ecosystem (Lüscher et al. 2000; Drissner et al. 2007).

Soil CO2 Efflux

Soil CO2 efflux showed a significantly higher Δ13C under eCO2 concentration (Fig. 4). To evaluate the actual amount of respired 13C under aCO2 and eCO2 concentrations, the isotopic values need to be weighted with the soil CO2 efflux. However, soil CO2 efflux was not assessed in this experiment, but it has been convincingly shown that eCO2 concentration increases soil CO2 efflux in grasslands (Luo et al. 1996; Selsted et al. 2012) and other ecosystems (Carney et al. 2007; Hyvönen et al. 2007). Therefore, the increased Δ13C of soil-respired CO2 under eCO2 conditions is probably also associated with a higher soil CO2 efflux and the respiration of recently assimilated 13C would therefore be higher under eCO2 concentration.

Soil respiration is mainly composed of root and microbial respiration. Increased root biomass in L. perenne (Hebeisen et al. 1997; Van Kessel et al. 2000a; Daepp, Nösberger & Lüscher 2001) under eCO2 concentration is a potential source of recently assimilated 13C to the soil CO2 efflux. Furthermore, increased symbiosis of N-fixing bacteria with T. repens (Schortemeyer et al. 1996; Zanetti et al. 1996; Montealegre et al. 2000) is another potential source of 13CO2 production. Also, a higher fungal importance under low nitrogen fertilization (Drissner et al. 2007) and an increased utilization of 13C compared at aCO2 can add a substantial amount of 13C to the soil CO2 efflux. CO2 originating from theses sources together can result in the observed higher Δ13C in soil respiration.

The main contribution of AMF to the measured Δ13C in soil CO2 efflux under eCO2 probably appeared within 16 h after the pulse labelling (Jakobsen & Rosendahl 1990) and was not captured. This assumption is in agreement with the finding that mycorrhizal fungi are strongly dependent on photosynthetic activity (Moyano, Kutsch & Schulze 2007) and a faster carbon turnover under eCO2 concentration would therefore result in an earlier 13C enrichment peak. This is supported by the observed lateral 13C transport in 15 cm distance from the labelled plots where the 13C enrichment peak in soil respiration occurred 1 day after labelling. This peak was delayed in the aCO2 compared to eCO2 treatment and can be explained by carbon transport within the mycorrhizal network that can cover relatively large distances (Johnson, Leake & Read 2002).

Soil Carbon Storage Potential

Plant-assimilated carbon can be distributed to different carbon pools, but will eventually be released back to the atmosphere. The increased carbon input to the soil under eCO2 led to increased soil microbial biomass (Sowerby et al. 2000; Drissner et al. 2007) but the soil-respired CO2 per unit microbial biomass stayed unchanged (Sowerby et al. 2000) likely due to higher rates of SOM mineralization (Drissner et al. 2007). After 4 years of eCO2 fumigation in the Swiss FACE, the soil carbon pool remained unchanged under eCO2 concentration as shown in previous studies (Van Kessel et al. 2000a,b). Our observations agree with those studies that reported a stimulated microbial community under eCO2 and high impact of the fungal group in the studied ecosystem. Our results suggest that eCO2 concentration increases the carbon throughput of the ecosystem (Figs 3 and 4) due to a higher soil carbon input and an increased soil microbial biomass. A higher carbon throughput but unchanged carbon allocation patterns under aCO2 and eCO2 concentrations point towards an unchanged net ecosystem carbon balance which is in agreement with previous observations (Van Kessel et al. 2000a,b).

Conclusions

We presented the fate of carbon into above-ground and below-ground carbon pools in mixed grass-clover vegetation in a 13CO2 pulse labelling experiment after a decade of CO2 fumigation. In this agricultural relevant vegetation mixture, gram-negative bacteria and fungi have been shown to be important for the nutrient balance under eCO2 especially. Our data suggest that the increase in microbial biomass is counterbalancing the increased plant nutrient demand under eCO2 concentration. Furthermore, the increase in below-ground carbon allocation is balanced by a higher microbial activity, and thus, the release of CO2 back to the atmosphere is increased. We conclude that, due to these balancing mechanisms, a decade of eCO2 concentration did not drastically change carbon allocation in the studied grass-clover ecosystem.

Acknowledgements

We thank the European Commission for awarding P. L. Staddon a Marie Curie Individual Fellowship. P. A. Olsson thanks the Swedish Research Council for financial support. We thank all the staff involved with the Swiss FACE experiment at ETH Zürich for making this work possible, especially I. Popa for help with sample collection. We also thank N. Ostle at CEH Merlewood for providing the labelling chambers and A. H. Fitter, J. P. Grime and K. Thompson for comments on an earlier version of this paper.

Ancillary