Isotopic methods for non‐destructive assessment of carbon dynamics in shrublands under long‐term climate change manipulation

Long‐term climate change experiments are extremely valuable for studying ecosystem responses to environmental change. Examination of the vegetation and the soil should be non‐destructive to guarantee long‐term research. In this paper, we review field methods using isotope techniques for assessing carbon dynamics in the plant–soil–air continuum, based on recent field experience and examples from a European climate change manipulation network. Eight European semi‐natural shrubland ecosystems were exposed to warming and drought manipulations. One field site was additionally exposed to elevated atmospheric CO2. We discuss the isotope methods that were used across the network to evaluate carbon fluxes and ecosystem responses, including: (1) analysis of the naturally rare isotopes of carbon (13C and 14C) and nitrogen (15N); (2) use of in situ pulse labelling with 13CO2, soil injections of 13C‐ and 15N‐enriched substrates, or continuous labelling by free air carbon dioxide enrichment (FACE) and (3) manipulation of isotopic composition of soil substrates (14C) in laboratory‐based studies. The natural 14C signature of soil respiration gave insight into a possible long‐term shift in the partitioning between the decomposition of young and old soil carbon sources. Contrastingly, the stable isotopes 13C and 15N were used for shorter‐term processes, as the residence time in a certain compartment of the stable isotope label signal is limited. The use of labelled carbon‐compounds to study carbon mineralisation by soil micro‐organisms enabled to determine the long‐term effect of climate change on microbial carbon uptake kinetics and turnover. Based on the experience with the experimental work, we provide recommendations for the application of the reviewed methods to study carbon fluxes in the plant–soil–air continuum in climate change experiments. 13C‐labelling techniques exert minimal physical disturbances, however, the dilution of the applied isotopic signal can be challenging. In addition, the contamination of the field site with excess 13C or 14C can be a problem for subsequent natural abundance (14C and 13C) or label studies. The use of slight changes in carbon and nitrogen natural abundance does not present problems related to potential dilution or contamination risks, but the usefulness depends on the fractionation rate of the studied processes.


| INTRODUCTION
Global climate change scenarios predict that increased greenhouse gas (e.g. CO 2 , CH 4 and N 2 O) concentrations in the atmosphere will alter the periodicity and magnitude of drought events and will increase mean global temperatures by approximately 0.2°C per decade (IPCC, 2013). For the European continent this will manifest as drier summers in the South and increased precipitation in the North (IPCC, 2013). Elucidating the consequences of such atmospheric changes for biogenic carbon fluxes is one of the main challenges for the scientific community. Some models have predicted a positive feedback to climate change, resulting from higher increases in respiratory fluxes from ecosystems (e.g. carbon release through soil respiration) than in net primary productivity, which would lead to further increases in atmospheric CO 2 (Denman et al., 2007;Friedlingstein et al., 2006). To assess the likelihood of this positive feedback, experimental studies that analyse the long-term adaptations of ecosystem carbon fluxes to climate change are critically needed. However, climate change experiments are often conducted at short or medium timescales due to funding constraints, or due to the limited life span of the experimental plots, as repeated removal of samples often leads to disturbances and experimental artefacts in the studied system. Hence, there is a necessity for the maintenance of long-term experiments using nondestructive methods.
Carbon fluxes through the plant-soil-air continuum play a central role in soil carbon cycling (Phillips, Fox, & Six, 2006;Zak, Pregitzer, King, & Holmes, 2000). Consequently, above-ground to below-ground fluxes might largely determine carbon emissions from ecosystems under the different climate change scenarios (Chapin et al., 2009).
Stable carbon isotope studies can give important insights into carbon fluxes through the plant-soil-air continuum with the minimal disturbance to the system. The isotopic carbon composition of compartments in this continuum is a result of the different isotope fractionation processes along the pathway from CO 2 fixation by plants to carbon allocation to soil (reviewed in Brüggemann et al., 2011). Thus, the analysis of the natural abundance of carbon isotopes in these compartments can give information about some processes related to photosynthesis and carbon losses through plant or soil respiration. In addition, in situ pulse labelling with the heavy stable carbon isotope ( 13 C) is a powerful tool to analyse short-term dynamics of carbon allocation to the soil with high resolution (Epron et al., 2012;Högberg et al., 2008;Reinsch & Ambus, 2013). The application of these isotopic methods can therefore provide unique information about aboveground-below-ground linkages and their alterations in response to climate changes.
In order to investigate long-term effects of climate change on shrubland ecosystems, an experimental network was established across Europe (the INCREASE network). Studying the response of shrublands to climate change is important, since they are representative ecosystems in Mediterranean and North European countries, where they play an important ecological role in preserving biodiversity (Wessel et al., 2004). In addition, land area covered by shrublands has dramatically decreased in temperate Europe during the past century, due to land use changes, increased pollution and eutrophication, and climate change (Fagúndez, 2013). In Mediterranean regions, however, shrublands have increased their extension due to land abandonment (Fagúndez, 2013).
Within the climate change network, common non-destructive methods were used across sites to ensure the comparison of treatment effects across different climatic regions. Evaluating the impact of climate change treatments on shrubland carbon dynamics was one of the main objectives of this experimental network, and thus a range of methodologies to quantify and trace distinct carbon pools and their fluxes have been applied since 1999. Priority was given to those techniques that minimise disturbances to vegetation and soil to guarantee long-term research.
Here, we review isotope methods that have been applied across this climate change experimental network to study ecosystem carbon mineralisation by soil micro-organisms enabled to determine the long-term effect of climate change on microbial carbon uptake kinetics and turnover.
4. Based on the experience with the experimental work, we provide recommendations for the application of the reviewed methods to study carbon fluxes in the plantsoil-air continuum in climate change experiments. 13 C-labelling techniques exert minimal physical disturbances, however, the dilution of the applied isotopic signal can be challenging. In addition, the contamination of the field site with excess 13 C or 14 C can be a problem for subsequent natural abundance ( 14 C and 13 C) or label studies. The use of slight changes in carbon and nitrogen natural abundance does not present problems related to potential dilution or contamination risks, but the usefulness depends on the fractionation rate of the studied processes.

K E Y W O R D S
14 C, bomb-C, drought, free air carbon dioxide enrichment, pulse-labelling, stable isotopes, warming dynamics in the plant-soil-air continuum. In particular, we focus on methodologies that: (1) analyse the abundance of naturally rare isotopes of carbon ( 13 C and 14 C) and nitrogen ( 15 N) in the different ecosystem compartments, (2) trace experimentally-induced changes in the isotopic signatures to assess rhizodeposition utilisation by soil biota, and (3) manipulate and trace the isotopic composition of C-compounds to analyse C mineralisation by soil micro-organisms in laboratory studies. Along-side the methods, data from the field studies are presented as accompanying illustrative boxes, and practical recommendations for the applications of these methodologies at large-scale climate change experiments are outlined in Table 1.

| THE EXPERIMENTAL CLIMATE CHANGE NETWORK INCREASE
The experimental network for the study of climate change impacts on European shrublands (INCREASE, "An Integrated Network on Climate Research Activities on Shrubland Ecosystems") was established in 1998. The network is comprised of eight shrublands situated across a natural temperature gradient of mean annual temperature from c.
8°C in the North to c. 16°C in the South, and a rainfall gradient ranging from 510 mm to 1,741 mm from East to West (Figure 1). These sites represent Continental, Atlantic and Mediterranean shrublands.
At each site, whole-ecosystem warming and drought treatments were applied in triplicates of 20 m 2 plots, using automated retractable curtain constructions (see Beier et al., 2004;Mikkelsen et al., 2008 for a full description). At one of the Danish sites (DK-BRA), a FACE treatment was installed, and combinations of the climate treatments were established and resulted in a plot size of 9 m 2 . Climatic conditions at the plot level (air temperature, humidity, soil temperature and moisture) were recorded in half-hour or hourly intervals, and main carbon pools and fluxes have been periodically monitored, including aboveground plant biomass (Kröel-Dulay et al., 2015), litter production, soil respiration and net ecosystem carbon exchange Lellei-Kovács et al., 2016).

| Ecosystem processes reflected by stable isotope fractionation ( 13 C and 15 N)
The relative abundance of the rare and heavy stable isotopes of nitrogen ( 15 N) and carbon ( 13 C) compared to the most abundant stable isotope, 14 N and 12 C, respectively, is expressed as the delta (δ) notation (e.g. δ 13 C and δ 15 N in ‰), which is the deviation of the 13 C or 15 N abundance in the sample compared to a reference material (Brand & Coplen, 2014). Most natural processes (chemical, physical or enzymatically catalysed) discriminate against heavy isotopes (e.g. 13 C, 15 N, 18 O), which in open systems results in an isotopically depleted product with comparably smaller concentration of the heavy isotopes than its corresponding substrate (Fry, 2006). If the dominant process rate changes, or if the substrate is exhausted, then the δ value of the T A B L E 1 Suggestions and advice to consider when applying isotopic methods for the study of carbon fluxes in the plant-soil system

Method Expenses (cost) Advice (do's and don'ts)
Bomb-C (natural 14 C abundance) High (AMS analysis); Equipment for CO 2 sampling is cheap (closed chambers, carbon-free pump, batteries, and molecular sieve system). An IRGA is also required Avoid materials and labs with possible 14 C contamination. If soil CO 2 is to be analysed in the field, long incubation times are required to get sufficient CO 2 for AMS analysis (typically >1 ml). Think carefully about the soil depths to be analysed, and take the sample consistently. 14 C signatures might vary strongly along few cm in the soil If bulk soil 14 C is to be analysed, try to remove the roots as much as possible, because of their contrasted 14 C signature In situ 13 C-CO 2 pulse-labelling 13 C-enriched compounds used for labelling and as standards are usually expensive; 13 C determination in specific compounds is expensive, although cheaper than AMS Consider the target pools to be analysed and the potential dilution of the label by the unlabelled root system or soil carbon pool If your study requires a high 13 C enrichment, mind the potential risk of contaminating the site Avoid above ambient CO 2 concentrations in the chamber If you need to monitor CO 2 during your pulse, remember that IRGAs are rather insensitive to 13 CO 2 Natural abundance of isotopes ( 13 C and 15 N) IRMS analysis is relatively cheap Make sure the history of sampling site is known (previous labelling experiments?)

C-substrates mineralisation
Analysis of the trapped 14 C-CO 2 is relatively cheap High risk of contaminating laboratory equipment 13 C-injection in situ Similar to 13 C-CO 2 pulse-labelling Contamination risk of 13 C leaching is present, but smaller to our judgement than from 13 C-CO 2 experiments Do not use areas dedicated to natural abundance work product (such as the plant leaf) may significantly change, due to the underlying fractionation.
Decreases in soil water availability due to drought can alter the isotope signature of both carbon and nitrogen in the above-ground plant biomass. During drought stress, leaves reduce stomatal opening to preserve water. As this happens, the space that confines the air as an immediate source of CO 2 for photosynthesis (the sub-stomatal cavity) becomes a more closed system due to the restriction of the renewal of CO 2 , and as a result a higher proportion of the heavy 13 C in CO 2 is fixed by Rubisco (C3 plants; Tcherkez, Mahe, & Hodges, 2011).
Hereby the discrimination against the heavy 13 C isotope is decreased.
As a consequence, in plants with a C3 photosynthetic pathway a 13 C enrichment in the leaf occurs during drought stress (Cernusak et al., 2013). Indeed, the 13 C enrichment at the leaf level is related to an increased intrinsic water use efficiency (WUEi), the ratio of assimilation to stomatal conductance (Farquhar & Richards, 1984). Changes in soil water availability may also alter the leaf nitrogen isotope signature by changing the nitrogen availability with soil depth, and thereby the 15 N signature of the plant nitrogen source (Lloret, Peñuelas, & Ogaya, 2004). In general, an increase in the δ 15 N signature in the leaves indicates a progressive N saturation and/or N losses in the surrounding system because all major pathways of N loss (denitrification, ammonia volatilisation and nitrate leaching) cause δ 15 N enrichment of the remaining nitrogen (Peñuelas, Filella, Lloret, Piñol, & Siscart, 2000).
Interpretation of changes in leaf δ 15 N, however, is not straightforward since leaf δ 15 N signatures might largely depend on mycorrhizal associations, and shifts in nitrogen sources between organic and inorganic compounds under a drought or warming could influence the leaf δ 15 N as well (Andresen, Michelsen, Jonasson, Beier, & Ambus, 2009;Michelsen, Quarmby, Sleep, & Jonasson, 1998). For instance, the increase in plant δ 15 N values with aridity may also result from increasing reliance on recycled organic N sources as opposed to new inputs.
Across the INCREASE network the effects of warming and drought on plant 13 C and 15 N natural abundance was monitored over four years, starting two years after onset of the climate manipulation. Current year shoots or leaves were analysed for δ 13 C and δ 15 N immediately after each artificially prolonged drought. Plant material was dried at 70°C and ground to a fine powder before analysis of δ 13 C and δ 15 N using isotope ratio mass spectrometry (IRMS).
We expected to find higher δ 13 C values: (1) in drought treated plants (compared to control plots) and, (2) in plants growing at drier locations across the precipitation gradient (for a given common plant species). Furthermore, we expected (3) the δ 15 N to change in response to drought, as the nitrogen source (depth) is changed (at one location, within-species). Some significant effects of the drought treatment were observed on plant tissue δ 13 C and δ 15 N (Box 1).
Differences between years (effect of time) were more pronounced than the effect of the drought treatment for Populus alba δ 13 C (HU), Erica multiflora δ 15 N (SP) and Globularium alypum δ 15 N and δ 13 C (SP).
Only Calluna vulgaris showed a significant response to the drought treatment for δ 13 C as hypothesised (Box 1a). For C. vulgaris, which was growing at several locations (UK-CL, NL and DK-MOLS), the δ 13 C was higher at drier locations, when compared along the precipitation gradient, and also higher in the drought treatment at the

Before you start Data analysis hint Time spent
If you are not sure about potential 14 C contamination in your laboratory, use another laboratory or make a swipe test Make previous trials to assess the incubation times required to get a sufficient CO 2 sample Go through the whole process of sample preparation with a trial sample Discuss your results with the Radiocarbon facility staff Processing time depends on the type of sample, although is usually low; determination by AMS may take several months depending on the facility Test your chamber and tubing materials for adsorption/ desorption effects, and ensure these are without carbon content (use PTFE (Teflon) tape, not gluing paper-based) Make a previous trial if possible and go through the whole process of sample preparation Report the label addition per area: g 13 C m −2 Pulse labelling experiments are usually short, but intensive (high sampling frequency immediately after the pulse) Experiments requiring root washing or microbial compound extraction are time consuming Be aware that FACE can dilute the isotopic signal, most CO 2 enriched systems use 13 C depleted sources, because this is cheaper Sampling time and grinding / weighing of sample Analysis usually done at a dedicated natural abundance facility.
You need to work in a dedicated 14 C laboratory safely away from the natural abundance facility Continue sampling until decline in emission is level, this ensures better model fit Labelling intended for soil microbial components is more intense from 13 C liquid substrate in-situ injection than from 13 C-CO 2 pulse labelling Soil sampling is destructive, consider to have several parallel plots to harvest an undisturbed plot at each sampling event Sample handling from field work until the extraction takes a few days so plan only one sampling event per week if possible NL and UK-CL sites (Box 1b). Finally, we found no response of leaf δ 15 N to drought or warming, however, P. alba had a much depleted δ 15 N relative to the other species. We attribute these differences to species specific utilisation of different nitrogen sources (perhaps more dependent on nitrate at the HU site) or different mycorrhizal associations with higher rates of isotopic fractionation.

| Bomb-14 C technique to asses sources of soil respiration
The natural radioactive 14 C abundance can be used to identify different sources of carbon in a mixed pool, for instance, in soil respiration. Radiocarbon signatures of more recent (i.e. <65-70 years) and older carbon sources are different as a result of the nuclear bomb tests in the atmosphere during the 1950/60s. These tests led to an increase in the 14 C content in the atmospheric CO 2 in the Northern hemisphere, which reached its maximum in 1963 ("bomb peak" doubling at c. 200% pMC). Ever since the subsequent atmospheric nuclear test moratorium, the "bomb-14 C" content has decreased due to the dilution with fossil fuel-derived CO 2 in the atmosphere and its incorporation in ocean and terrestrial carbon pools (Trumbore, 2009). Through its incorporation in plant biomass, the radiocarbon analysis of ecosystem fluxes found to contain bomb-14 C provides singularly unique information which crucially and directly confirms the "recent" origin of SP method (Cisneros-Dozal, Trumbore, & Hanson, 2006;Schuur & Trumbore, 2006;Subke, Voke, Leronni, Garnett, & Ineson, 2011).
In these studies, analysis of the 14 C-CO 2 signatures of roots and SOM was performed under controlled conditions and collated with analyses of field gas efflux (the mixed pool). Radiocarbon analysis of soil or ecosystem respiration has been used to evaluate the response of a range of ecosystems to different factors of climate change, such as increasing temperatures, decreasing rainfall or permafrost thaw (Borken, Savage, Davidson, & Trumbore, 2006;Muhr, Borken, & Matzner, 2009;Schuur et al., 2009). The method allows for a direct evaluation about possible differential effects of climate change factors on the fate of recent vs. older soil C moieties, a central question for climate change scientists.
The applicability, sensitivity and accuracy of the method is obviously improved when more of the "bomb-14 C" is detectable in the specific analysed C pool, e.g. containing relatively more C which laid down in living tissues and subsequent decomposition products in the 1950 to 1970s period.
We tested the effect of experimental warming and drought on the natural abundance of 14 C in respired soil CO 2 at early stages of the climate manipulations at the Peaknaze field site (UK-PK). Our hypothesis was that drought increased heterotrophic respiration more than warming in this seasonally waterlogged soil, due to a greater responsiveness of old soil carbon to drought relative to temperature as a driver (Domínguez, Holthof, Smith, Koller, & Emmett, 2017;Domínguez et al., 2015). Therefore, we expected the greatest 14 C-enrichment in the field-collected soil respiration samples from the drought plots. Soil efflux samples were collected in the late experimental drought period (September 2011), using a molecular sieve sampling system (Bol & Harkness, 1995;Hardie, Garnett, Fallick, Rowland, & Ostle, 2005) attached to closed dark respiration chambers placed on the soil overnight. CO 2 was subsequently recovered from the molecular sieve traps

| 13 C-CO 2 pulse labelling
In situ pulse labelling with the stable carbon isotope ( 13 C) is a good method to address questions related to the time-lag between carbon assimilation and CO 2 release from soil (Kuzyakov & Gavrichkova, 2010). In 13 C-CO 2 pulse labelling experiments, 13 C Repeated pulses of 13 C-CO 2 (99 atom% 13 C = 99% 13 C + 1% 12 C) were applied over 8 hr (Box 3). The domes were sealed to a frame which was inserted into the ground at least ten days before the pulse, and had several sealed septa to collect gas samples to estimate the concentration of the 13 C-labelled CO 2 . Plant leaves and soil from the rooting zone were collected at different times after the labelling, using a higher sampling frequency during the first hours after the pulse. Soils were freeze-dried, sieved to ≤ 0.05 mm and PLFAs were extracted. Fatty acid methyl esters (FAMEs) were analysed by gas chromatography combustion-isotope ratio mass spectrometry (GC-c-IRMS). The main challenge was the low recovery of 13 C label in the below-ground compartment, especially in individual developed experimental setup for pulse-labelling aimed (1) to be easily deployable in remote areas, (2) to distribute labelled 13 C-CO 2 to as many plots at the same time as possible to ensure similar and constant conditions for CO 2 uptake by the vegetation, and (3) to ensure constant CO 2 concentration available to the vegetation throughout the labelling period. Therefore, a mobile flow-through system suitable for continuous 13 C-CO 2 delivery was developed (Box 3): A gas-tight vinyl balloon (c. 3 m diameter) was filled with CO 2 free synthetic air and mixed with 13 C-CO 2 (50 or 99 atom%) that supplied the transparent chambers enclosing the vegetation of interest with air over the duration of the experiments, ranging from 4 to 7.5 hr. Air was pumped continuously through gas tight tubing via electric diaphragm pumps (Reinsch & Ambus, 2013). The first experiment was conducted at the end of the growing season (October 2010), when we observed the highest allocation of carbon below-ground as measured by 13 C in soil respiration (Reinsch et al., 2014). The second experiment was conducted in the spring (May 2011) and showed a major allocation of carbon to above-ground structures under elevated atmospheric CO 2 concentration, but carbon allocation to below-ground structures was higher in drought plots than in untreated control plots. The allocation of recently-assimilated carbon under warming conditions was similar to that under ambient conditions. The last experiment, BOX 2 Impact of warming and drought on the 14 C signature of soil respiration. (a) Records of atmospheric 14 C over the 20th century. The unit for 14 C signature (% Modern) is a measurement of the deviation of the 14 C/ 12 C ratio of a sample from the "Modern" standard, which is defined as 95% of the radiocarbon concentration (in AD 1950) of a reference material (NBS Oxalic Acid I, SRM 4990B), adjusted to a δ 13 C reference value of −19‰. (b) At the UK-PK site, the 14 C signature of the soil efflux was measured (bars, left axis). 14 C values were highly heterogeneous (ranging from 105.49 to 110.13% Modern; values of >100% Modern suggest that a substantial component -and potentially all-of the carbon was trapped by photosynthesis during the post-bomb era, that is, since ~AD 1955).There were no significant effects of the warming treatment, while there was a trend towards the release of older carbon in the drought plots. On average, the carbon being released from the plots had been fixed from the atmosphere between six and eight years earlier (line, right axis). (c) Detail of a closed static chamber used to collect CO 2 from the soil efflux BOX 3 Analysis of rhizodeposit utilisation by microbes using in-situ 13 C-CO 2 pulse-labelling experiments. (a) At the Clocaenog site (UK-CL) this technique was applied along a peat layer gradient. Repeated pulses of 13 C-CO 2 were applied during 8 hr to C. vulgaris using sealed transparent domes attached to a core inserted into the ground. (b) The incorporation of 13 C into soil microbial PLFAs was analysed. Despite a high applied dose of 13 C (99 atom%), the dilution of the tracer within the large pool of unlabelled root biomass was remarkable, and as a consequence most of the analysed PLFAs showed no 13 C enrichment. (c) 13 C recovery in Gram-negative bacteria after a 13 C-CO 2 pulse at the Brandbjerg site (DK-BRA). The enrichment pattern in PLFAs attributed to Gram-negative bacteria in soils exposed to drought and elevated CO 2 concentration (+120 ppm) for 8 years show different carbon utilisation patterns and magnitudes under imposed climatic conditions implying changed carbon cycle dynamics. (d) Flow-through pulse-labelling equipment showing the gas reservoir containing 13 C-CO 2 for up to 8 hr of labelling connected to transparent Plexiglas chambers via tubing (a) iC 1 4 : 0 C 1 4 : 0 i-C 1 5 : 0 a -C 1 5 : 0 C 1 5 : 0 i-C 1 6 : 0 C 1 6 : 0 C 1 6 : 1 '9 c C 1 6 : 1 '1 1 c C 1 7 : 0 D 9 , 1 0 C 1 8 : 0 C 1 8 : 1 '9 c C 1 8 : 1 '1 1 c C 1 8 : 2 ( n -6 ) C 1 8 : 3 ( n -3 ) conducted in early season 2013 (June), was performed during a period with impeded photosynthetic activity and indicated that labelling performance is poor when vegetation is recovering from harsh winter conditions with bare frost or severe drought conditions (Box 3).
Thus, it is important that the vegetation of interest displays green, photosynthetically active structures to facilitate CO 2 uptake and sufficient labelling of ecosystem carbon pools. From these labelling experiments we learned that climate change factors change the flow of carbon within the plant-soil-atmosphere continuum. Increased atmospheric CO 2 concentrations accelerate the carbon cycle as seen as labelled carbon through the bacterial community over time. In contrast, drought slowed down carbon transport dynamics with soil microbes showing the 13 C label later in time (Reinsch et al., 2014).
Our studies illustrate the complexity of controlling in situ pulselabelling experiments in ecosystems dominated by woody plants, which is even more challenging with 13 C-CO 2 than with 14 C-CO 2 because of their respective atmospheric backgrounds and detection limits (Epron et al., 2012). Ideally, 13 C doses for in situ use should be carefully tested in trials, considering the nature of the studied vegetation and the compounds to be analysed. If for example, specific compounds of the soil microbial biomass are the main interest, then strong isotopic doses should be applied, and it is advisable to deploy the 13 C pulse when plants naturally allocate carbon below-ground, for example, when preparing for winter. The 13 C signal can be increased, using highly labelled 13 C-CO 2 (99 atom %). However, the usage of a highly enriched CO 2 can potentially lead to blurry signals and has to be applied with caution (Watzinger, 2015). Furthermore, 13 C-CO 2 concentration inside the labelling chamber should be as close as possible to ambient values, because unrealistic high CO 2 concentration will change plant CO 2 uptake. Repeated moderated 13 C-CO 2 applications during longer exposure times might be more appropriate, but inside closed transparent chambers temperature and humidity may increase if the labelling period is prolonged, which also affects photosynthetic processes (Epron et al., 2012). Losses of 13 C due to physical diffusion and adsorption/desorption into the chamber and tubing material should also be considered. In particular, the back-diffusion of the 13 CO 2 from the soil to the atmosphere which entered the soil pores during the labelling might confound the interpretation of measured below-ground respiration (Selsted et al., 2011;Subke et al., 2009).
However, when applied properly, the insights into terrestrial carbon allocation can be detailed and novel (Box 3).

| Free air carbon dioxide enrichment (FACE)labelling
An alternative method for 13 C labelling of vegetation and wholeecosystems is to use 13 C-depleted CO 2 in FACE experiments. The FACE technique has through decades been used within cropping systems (Kimball, 2016), grasslands (Hovenden, Newton, & Wills, 2014;Mueller et al., 2016;Reich, Hobbie, & Lee, 2014) and forests (Terrer, Vicca, Hungate, Phillips, & Prentice, 2016) experiments, with the primary goal of assessing potential carbon dynamics and enhancement of plant growth (Andresen et al., 2016). As a side effect, the change in carbon isotopic composition of vegetation exposed to the FACEtreatment can be used to trace freshly assimilated carbon into soil microbial biomass, fauna and organic carbon pools. This approach was used at the Brandbjerg site (DK-BRA). The CO 2 used to elevate concentrations of atmospheric CO 2 to 510 ppm had δ 13 C values ranging from −3.0 to −36.7‰ throughout 8 years of experimental treatment, with an overall mean of −26.1‰. The source of the CO 2 supplied was brewery surplus CO 2 as a chemically obtained side product. The mixing of the added CO 2 via FACE with ambient CO 2 in the moving air mass resulted in a 13 C depletion ranging from −6.7 to −15.6‰. On average, this equals a depletion of CO 2 in FACE plots of −4.8‰ relative to the atmospheric −8‰ average. Ecosystem carbon pools became depleted accordingly, and the FACE-13 C depletion acted as a long-term persistent isotope labelling. As a result, soil fauna (Enchytraeids) sampled from each of the climate-treated plots was significantly depleted in δ 13 C by −0.5 to −2.0‰ in the CO 2 treatments (Andresen et al., 2011). This was due to translocated 13 C substrate through the food web, starting with plant assimilation of 13 C-depleted CO 2 , followed by plant root exudation and microbial utilisation of the 13 C-depleted substrate and eventual digestion of microbes by enchytraeids. Hereby the freshly supplied carbon source was recognised to be transferred in the natural setting, within a given timescale. Also microbial biomass and PLFAs had a different baseline of 13 C content in ambient (nottreated) plots compared to CO 2 treated plots (Andresen et al., 2014).
This was used for the calculation of 13 C enrichments in each PLFA biomarker, also illustrating the pathway of newly assimilated carbon into microbial biomass.
A general drawback of the 13 C-FACE label is again the contamination of the surroundings, as even short and small un-planned draft winds can carry the depleted label onto "ambient" plots, and these will most likely be "contaminated" with 13 C (though not markedly exposed to high CO 2 concentrations) after some years of FACE activity. Therefore, one needs to collect reference material for the "natural abundance" level well away from the FACE experiment. Also, FACE-CO 2 can only be used as tracer if the isotopic composition of the FACE-CO 2 is considerably different than the isotopic composition of the atmospheric CO 2 .

| In situ injection of 13 C-enriched substrate solutions
As a much more localised approach, in situ injection of 13 C-and 15 N-enriched substrates directly below the soil surface can be used to assess the competition for the substrate between (1) plants and soil microbes, (2) microbial groups, and (3) the effects of the climate change treatments upon the competition for carbon or nitrogen substrates. Much research has focused on the sharing of nitrogen sources between plant and microbes (Kuzyakov & Xu, 2013), using in situ soil injections of 15 N labelled inorganic nitrogen (ammonium and nitrate) or organic nitrogen (amino acids) (Sorensen, Michelsen, & Jonasson, 2008). Once amino acids with dual labelled compounds ( 15 N and 13 C) were available for experimental use, double-labelled substrate was used to explore, for example, plant uptake of intact amino acids (Näsholm, Kielland, & Ganeteg, 2009;Rasmussen, Sauheitl, Eriksen, & Kuzyakov, 2010), and microbial utilisation of carbon substrates (Dungait et al., 2013;Rinnan & Baaaath, 2009).
In a labelling experiment at the DK-BRA site, amino acid injections into the soil were conducted to analyse the impact of the climate treatments on the uptake of free amino acid nitrogen by plants and soil microbes. Dual-labelled glycine ( 13 C 2 15 N-glycine: 99 atom% 13 Cof both carbon atoms-and 99 atom% 15 N) was added to 20 × 20 cm 2 subplots (Andresen et al., 2009). Each sub-plot received 0.1 L of redemineralised water labelled with 0.027 g glycine, corresponding to 687 mg glycine m −2 (223 mg C m −2 or 0.016 mg glycine g −1 dry weight soil). The label was injected into the soil just below the soil surface with a syringe moved among 16 evenly spaced points of a template, placed on top of the vegetation (Andresen et al., 2009). One day (c. 24 hr) after labelling with glycine, soil cores were sampled from the soil surface to 15 cm depth, for determining the relative uptake of the amino acid in plant roots (IRMS solid sample) and soil microbes. As in many other soil labelling experiments, the largest label recovery (measured by 15 N recovery since respiratory losses of 13 C remain unknown) was found in the total microbial biomass compared to total plant biomass (Kuzyakov & Xu, 2013). A subsample of fresh soil was extracted with re-demineralised water, and another set of subsamples was first vacuum-incubated with chloroform for 24 hr to release microbial carbon and nitrogen (Brookes, Landman, Pruden, & Jenkinson, 1985;Joergensen & Mueller, 1996), before extraction with re-demineralised water. A third subsample of soil was freeze-dried and later used for PLFA extractions. The 13 C enrichment in PLFA markers thus indicated the activity (vitality) of the specific microbial group (Watzinger, 2015).
We found that bacteria opportunistically utilised the freshly added glycine substrate, that is, incorporated 13 C, whereas fungi showed only minor or no glycine derived 13 C-enrichment (Andresen et al., 2014).
In comparison, 13 C traced into the microbial community via the 13 C-CO 2 pulse label at the same site (DK-BRA) also reached the bacterial community first. Bacteria showed high 13 C enrichment compared to fungal groups (Reinsch et al., 2014). This suggests that in situ injection of 13 C substrates might be a plausible alternative to mimic rhizodeposition effects. With the direct addition of 13 C label to the soil, a strong labelling of the microbial community was more easily achieved than with the indirect 13 C labelling of microbes via plant assimilated 13 C-CO 2 (Box 3).

| USE OF LABELLED CARBON-COMPOUNDS TO ANALYSE CARBON MINERALISATION BY SOIL MICROORGANISMS
Since soil micro-organisms have an important role in controlling the availability of nutrients via mineralisation of SOM, our understanding of how microbial functioning in the ecosystem is altered by global change must be improved (Grayston, Vaughan, & Jones, 1997).
These methods are all based on quantifying CO 2 respired during the mineralisation of organic carbon compounds that vary in size, charge and structural complexity. The first approach, Biolog MicroPlateTM (Biolog), assesses the catabolic diversity of soil organisms using a microtitre plate by incubating a soil culture in the presence of nutrients and 95 different carbon substrates; respired CO 2 is used to reduce a tetrazolium violet salt, which results in a colour change that can be quantified colorimetrically (Garland & Mills, 1991). This approach, however, has been criticised for bias towards fast growing organisms that thrive in culture (Preston-Mafham, Boddy, & Randerson, 2002).
In response to the criticism of the Biolog method, Degens and Harris (1997) developed a method based on SIR where individual substrates are added to intact soil and evolved CO 2 is sampled and quantified.
Finally, Campbell et al. (2003) combined aspects of both methods (MicroRespTM) where the response to carbon substrate addition to soil is measured colorimetrically, using a cresol red indicator dye in a microtitre plate format.
Community level physiological profiling of soil samples collected from all treatments across the network was conducted to determine the catabolic utilisation profile, turnover and pool allocation of low molecular weight (LMW) carbon compounds, using a selection of 14 Clabelled substrates. This method enabled the attribution of respired CO 2 to specific metabolic processes that facilitates the quantification and qualification of microbial mineralisation kinetics of substrates varying in structural complexity and recalcitrance. The kinetics of microbial 14 C-CO 2 evolution can be described using a first order exponential decay model (Box 4). The number of terms used in the exponential decay model can be used to explain how microbial kinetics relates time, substrate complexity and carbon pool allocation to, for example, rapidly cycled labile soil solution carbon, microbial structural carbon and recalcitrant extracellular soil organic carbon (Boddy, Hill, Farrar, & Jones, 2007;Kuzyakov & Demin, 1998;Nguyen & Guckert, 2001

| CONCLUSIONS AND RECOMMENDATIONS
Stable isotope studies provide insightful information about carbon dilute the 13 C signal. As an alternative, direct injection of 13 C-enriched substrates into soil can be applied to mimic rhizodeposition and to achieve a higher 13 C signal in the microbial community. Fumigation with FACE-CO 2 can be used to achieve a longer-term labelling of soil microbes and fauna.
The application of these techniques, however, is not exempt from difficulties and disadvantages. To keep a high caution and avoid mistakes, our collective recommendations for applying the described methods are provided and addressed in Table 1.
For in situ pulse-labelling studies there are major seasonality constraints to the distribution of the label throughout the ecosystem compartments, that is, the seasonality of carbon allocation below-ground due to changing plant activity, or the plant health status which determines the amount of tracer entering the system. Importantly, field plots previously "contaminated" by highly enriched isotope labelling should be considered potentially inoperable for further scientific isotope studies using the natural abundance approach. However, plant and soil structures remain largely undisturbed. In outlook for setting up a large-scale climate manipulation, areas that have not been previously used for experimental BOX 4 Exponential decay kinetics for 14 CO 2 evolution during microbial 14 C substrate mineralisation. The catabolic utilisation profile, turnover and pool allocation of low molecular weight (LMW) carbon substrates was determined in soils collected across the experimental network. Sixteen 14 C labelled amino acids and sugars varying in structural complexity and recalcitrance were used in a multiple substrate induced respiration (SIR) assay on soil. Evolved CO 2 was collected using NaOH traps and absorbed 14 CO 2 was measured with a scintillation counter. (a) For substrate mineralisation a doubleterm first order decay model with an asymptote fitted the data with an r 2 of .99. Using the coefficients from the fitted equation, estimated half-life of the substrate in the first phase (soil solution uptake) was 30 hr, and in the second slower phase (microbial turnover) 408 hr. Approximately, 40% of the substrate was immobilised in the soil, 48.3% respired during the first phase, and 13.2% respired during the slower second phase. (b) Half-life of the substrate in the soil solution vs. mean annual temperature, in control (triangle) and warming (circle) treatments, data points are M ± SE (n = 3). Warming treatment and relative warmer site, simply increases the catabolic utilisation of labile LMWcarbon until a threshold mean annual temperature of 11.5°C work with isotopes should consequently be selected. In particular, the "bomb-C" method is very sensitive to the contamination of soil or plant samples with 14 C-enriched material, and thus its application should be limited to sites and facilities where no 14 C-labelling work has been conducted. Additionally, it should be noted that any history of fertilisation might also alter the natural isotope abundance of ecosystem compartments (in particular 15 N signatures), potentially confounding experimental results. The surroundings of a FACE experiment can be also "contaminated" by draft winds carrying the depleted label onto ambient plots. Authors state that they do not have any conflicts of interest to declare.

AUTHORS' CONTRIBUTIONS
All authors contributed to the collection of the data included as illustration of the methodologies. L.C.A., M.T.D., S.R. and A.R.S. wrote the manuscript. All authors contributed critically to the drafts and gave final approval for publication.

DATA ACCESSIBILITY
Datasets associated with this article are available through Dryad Digital Repository https://doi.org/10.5061/dryad.mc460 .