Research Article Received: 5 May 2014
Revised: 11 August 2014
Accepted: 12 August 2014
Published online in Wiley Online Library
Rapid Commun. Mass Spectrom. 2014, 28, 2341–2351 (wileyonlinelibrary.com) DOI: 10.1002/rcm.7017
Minimising methodological biases to improve the accuracy of partitioning soil respiration using natural abundance 13C Helen S. K. Snell1,2*, David Robinson1 and Andrew J. Midwood2 1
Institute of Biological and Environmental Sciences, University of Aberdeen, Cruickshank Building, St Machar Drive, Aberdeen AB24 3UL, UK 2 Environmental and Biochemical Sciences, James Hutton Institute, Craigiebuckler, Aberdeen AB15 8QH, UK RATIONALE: Microbial degradation of soil organic matter (heterotrophic respiration) is a key determinant of net ecosystem exchange of carbon, but it is difficult to measure because the CO2 efflux from the soil surface is derived not only from heterotrophic respiration, but also from plant root and rhizosphere respiration (autotrophic). Partitioning total CO2 efflux can be achieved using the different natural abundance stable isotope ratios (δ13C) of root and soil CO2. Successful partitioning requires very accurate measurements of total soil efflux δ13CO2 and the δ13CO2 of the autotrophic and heterotrophic sources, which typically differ by just 2–8 ‰. METHODS: In Scottish moorland and grass mesocosm studies we systematically tested some of the most commonly used techniques in order to identify and minimise methodological errors. Typical partitioning methods are to sample the total soil-surface CO2 efflux using a chamber, then to sample CO2 from incubated soil-free roots and root-free soil. We investigated the effect of collar depth on chamber measurements of surface efflux δ13CO2 and the effect of incubation time on estimates of end-member δ13CO2. RESULTS: (1) a 5 cm increase in collar depth affects the measurement of surface efflux δ13CO2 by –1.5 ‰ and there are fundamental inconsistencies between modelled and measured biases; (2) the heterotrophic δ13CO2 changes by up to 4 ‰ within minutes of sampling; we recommend using regression to estimate the in situ δ13CO2 values; (3) autotrophic δ13CO2 measurements are reliable if root CO2 is sampled within an hour of excavation; (4) correction factors should be used to account for instrument drift of up to 3 ‰ and concentration-dependent non-linearity of CRDS (cavity ringdown spectroscopy) analysis. CONCLUSIONS: Methodological biases can lead to large inaccuracies in partitioning estimates. The utility of stable isotope partitioning of soil CO2 efflux will be enhanced by consensus on the optimum measurement protocols and by minimising disturbance, particularly during chamber measurements. Copyright © 2014 John Wiley & Sons, Ltd.
Quantifying CO2 emissions from soil organic matter (SOM) is essential for estimating a landscape’s carbon balance and for assessing the effects of climate, land-use and biodiversity on soil carbon stocks. Soil is the largest terrestrial store of carbon. The flux of CO2 from soils to the atmosphere is estimated at around 98 Pg of carbon per year,[1] the same order of magnitude as the flux attributable to primary productivity and about 10 times the amount that is due to anthropogenic emissions.[2] Carbon is lost from soil via the microbial respiration of SOM (heterotrophic) and added via the addition of plant biomass. Some of the CO2 assimilated photosynthetically by plants is released as CO2 during root respiration (autotrophic) and diffuses out of soil alongside the CO2 produced from SOM. Accurate assessment of the net ecosystem exchange of carbon (NEE, i.e. whether a landscape is a source or sink for CO2) relies on partitioning these autotrophic (RA) and heterotrophic (RH) contributions to total soil CO2 efflux (RS).
Rapid Commun. Mass Spectrom. 2014, 28, 2341–2351
Copyright © 2014 John Wiley & Sons, Ltd.
2341
* Correspondence to: H. S. K. Snell, Institute of Biological and Environmental Sciences, University of Aberdeen, Cruickshank Building, St Machar Drive, Aberdeen AB24 3UL, UK. E-mail: [email protected]
Methods to partition soil respiration often physically remove the autotrophic component in an experimental plot and infer the autotrophic contribution by comparison with an unaltered control plot. Such field techniques have been reviewed in detail elsewhere[3–6] and include root removal, trenching (to sever existing roots and prevent ingrowth of new ones) and stem girdling (whereby a ring of phloem is removed in order to halt the supply of photosynthate to the roots). However, SOM and root respiration are intrinsically linked by the priming effect, whereby the rate of microbial respiration of SOM (RH) is modified by the presence of live roots, which affect the soil environment via their mycorrhizal associations, nutrient uptake, exudation and physical presence.[7] Plant roots also alter the bacterial community structure,[8] soil carbon pools,[6] and moisture and temperature regimes of soil,[9] all of which can influence microbial respiration.[8,10,11] A potentially important limitation of field techniques that remove or sever roots (or reduce root respiration) is that they will also remove this primed component of heterotrophic respiration. This would result in an estimate of heterotrophic respiration that may not be an accurate reflection of the undisturbed flux. Natural abundance 13C offers an alternative means of partitioning the total soil-surface CO2 efflux (RS) and has the advantage that it requires no prior disturbance or alteration
H. S. K. Snell, D. Robinson and A. J. Midwood of the site. Briefly, this method relies on measuring the 13C content of RS and then of the CO2 respired from roots and root-free soil. A simple linear mixing model is then used to apportion the total flux into its RA and RH components. RS is typically sampled by placing a chamber on the soil surface. Chambers may be closed or open in design and sampled either discretely or continuously. Open, dynamic chambers, in which the headspace is supplied with CO2-free air (so that after a period of equilibration or flushing the CO2 in the chamber is derived only from soil-surface efflux and does not contain any atmospheric CO2), are useful as they minimise the error in measurement of δ13CRS.[12] Following sampling of the surface efflux, separated roots and root-free soil are incubated in vials or flasks,[9,13–15] or in gastight bags,[16–19] from which atmospheric CO2 is removed so that CO2 derived entirely from the root or soil respiration can accumulate. These three CO2 samples are analysed for their 13C/12C composition (δ13C) and the heterotrophic flux (RH) is calculated using the following equations: f RH ¼
δ13 CRS δ13 CRA δ13 CRH δ13 CRA
and f RA ¼ 1 f RH
2342
where fRH and fRA are, respectively, the fractions of total soil-surface efflux attributable to heterotrophic and autotrophic respiration; δ13CRS, δ13CRA and δ13CRH are the isotopic signatures of the total soil-surface efflux, autotrophic (soil-free root) respiration and heterotrophic (root-free soil) respiration. Heterotrophic flux is the product of fRH and total soil-surface flux, RS. Most natural and semi-natural temperate ecosystems are dominated by plants that use C3 photosynthesis and consequently the accumulated SOM is derived entirely from C3 vegetation. Microbial processing of dead plant material leads to a slight 13C-enrichment of the remaining SOM and as a result the difference between the isotope ratios of RA and RH is typically 2–8 ‰.[9,16,17,20] This range, although typical for natural abundance carbon isotopes, is small. Therefore, the integrity of isotopic partitioning of the CO2 flux depends heavily on the accuracy and precision of the δ13CO2 measurements of RS, RA and RH.[21] In some agricultural and naturally transitional ecosystems, C4 plants occur on historically C3 soil. Just as when plants are artificially labelled with 13C-depleted CO2, the difference between the RH and RA end-member δ13CO2 is greater in these cases and stable isotope partitioning estimates will probably be more accurate. However, these systems are uncommon, particularly in natural, temperate landscapes. Stable isotope partitioning studies report very varied endmember δ13CO2 values and source contributions. While some of this variability undoubtedly reflects the natural differences between ecosystems, some of it may well be due to methodological differences between studies. Coupled with the fact that stable isotope partitioning is not always successful (i.e. if the surface efflux δ13CO2 falls outside the range of the end members),[16] there is clearly scope to improve the utility of this approach by refining the methodology. Methodological biases can affect all three of
wileyonlinelibrary.com/journal/rcm
the δ13CO2 measurements required for stable isotope partitioning and are mainly associated with differences in chamber design and collar usage, and with disturbance effects during the separation of roots and soil. Chamber measurements of soil-surface efflux δ13CO2 are not straightforward due to the way in which the different isotopologues of CO2 diffuse through the soil. While the δ13C of the soil-surface CO2 efflux is, of course, defined by that of the combined CO2 sources, the different molecular weights of 12CO2 and 13CO2 mean that diffusive fractionation causes the CO2 within the soil profile to be 13C-enriched by around 4.4 ‰ compared with that of the sources and efflux (this fractionation is in fact a function of the δ13C of the respired CO2 and the soil [CO2], and varies by around 0.5 ‰ from this value). Further 13C-enrichment occurs in the upper soil layers due to mixing with (–8 ‰) atmospheric air. The 13C-enrichment in the soil profile is thus 4.4 ‰ at depth and is greater (approaching an absolute value of -8 ‰) close to the soil surface.[22,23] The very presence of the chamber can disturb the diffusion profile and bias the δ13C value of the CO2 diffusing into the chamber. The δ13CO2 of the measured efflux can be affected by deployment time, non-ambient chamber pressure and chamber [CO2]. These factors vary with chamber design and are the reason why different chamber methodologies record different values of efflux δ13CO2, both theoretically and in practice.[12,24–26] Rigid collars installed in the soil surface are commonly used to site chambers when sampling soil-surface CO2 efflux and are intuitively thought necessary to prevent the lateral ingress of atmospheric CO2 to the chamber. However, these too can influence the measurement of δ13CO2. For example, when CO2 diffusion from soil is modelled, open dynamic chambers tend to equilibrate with δ13C of the CO2 at the base of the collar, rather than with that of the true surface efflux.[25] Although the impact of collar depth on the measured δ13CO2 has been considered theoretically, it has not been tested in practice, is not standardised and is often overlooked when results are reported. Measurements of end-member δ13CO2 are similarly prone to disturbance effects. The process of excavating soil and separating it from roots exposes labile carbon substrates, depleted in 13C, which are preferentially respired by soil microbes. This causes an immediate and progressive decrease in δ13CO2 during incubation.[19] Root respiration exhibits species-specific changes to δ13CO2 following excision.[27] Thus, the method of soil and root separation (sieving or hand sorting) as well as the duration of the incubation will potentially introduce methodological biases to the measured end-member values. Finally, soil respiration studies are increasingly using cavity ringdown spectroscopy (CRDS) for isotopic analysis. While these systems have the advantage of portability and ease of use, the precision and accuracy of the analyses need to be optimised by a combination of operational protocols and post-analysis data handling; an issue that has yet to receive much consideration in the literature. The aim of this study was to improve the accuracy of isotopic flux partitioning by identifying and minimising methodological errors associated with some commonly used techniques. We tested the effect of collar depth on chamber measurements of δ13CO2 using grass mesocosms in the laboratory. We expected that the chamber measurement
Copyright © 2014 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2014, 28, 2341–2351
Methods in partitioning soil respiration using natural abundance13C would reflect δ13CO2 at the base of the collar and that, due to mixing with atmospheric air in the upper soil, shallower collars would give more 13C-enriched measurements than deeper ones. Furthermore, we expected that chambers used without collars would accurately measure δ13C of the surface efflux CO2, which is more 13C-depleted than the subsurface CO2 at any depth. We also tested whether chambers used without collars in the field were subject to contamination with atmospheric air. We investigated measurement errors during end-member incubations in the field to establish whether CO2 samples taken soon after root and soil excavation were representative of the in situ δ13CO2. We devised and applied correction factors to CRDS analysis to account for analyser drift and non-linearity.
controls the [CO2] in the chambers as well as the air flow to the CRDS analyser; chambers were sampled sequentially for 10 min each, which is enough time for the CRDS to stabilise and give a few minutes (13 analyses) of continuous data at each sampling time. A cylinder of reference gas (530 μmol mol–1 CO2 in a balance of air)[29] was analysed by CRDS every day and all data corrected to isotope ratio mass spectrometric (IRMS) analysis of the same gas using a Thermo Delta PlusXP isotope ratio mass spectrometer interfaced to a Gas Bench II (both from Thermo, Bremen, Germany) and a PAL autosampler (CTC Analytics, Zwingen, Switzerland)). Means of between 351 and 559 (in one case 273) daytime measurements of the δ13CO2 from each location were compared with Anova in Minitab 16 (Minitab Inc., Coventry, UK).
EXPERIMENTAL
Field site
Collar depth
Rapid Commun. Mass Spectrom. 2014, 28, 2341–2351
Air ingress On five days between the 25th and 30th April 2013 we field tested whether chambers used without collars were subject to lateral air ingress causing a difference in δ13CO2. We modified two of four chambers by the addition of a 4 cm wide foot on the chamber edge. The underside of the foot was made of flexible closed cell foam; when the chamber was weighted, the foot made a seal across its whole surface, even on uneven ground. The diffusive path length through the soil, from the atmosphere outside to the headspace inside the chamber, was thus 4 cm, the same as would be created by using a collar 2 cm deep but without severing roots or penetrating the soil. We compared these modified chambers with unmodified chambers placed directly on the soil surface without collars. Wet sand was placed around the chamber edge or foot and a 3 kg weight was placed on top of the chamber to ensure a good seal with the ground. Sixteen measurements of surface efflux were made both with and without the modified chamber foot. A general linear model with time of measurement as a random factor was used to test for a difference in δ13CO2 values between the modified and unmodified chamber measurements. Heterotrophic end member The heterotrophically respired δ13CO2 changes rapidly during incubation because soil disturbance exposes labile, 13 C-depleted substrates that are preferentially respired.[19]
Copyright © 2014 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm
2343
To investigate the influence of collar depth on the measured δ13C of soil-respired CO2, eight monoliths of grassland soil (Countesswells association, from the JHI Aberdeen site) measuring 60 × 40 cm and approximately 20 cm deep were extracted intact and placed in plastic boxes of the same dimensions. In our experience this is close to the maximum size of monolith that can be removed intact at this site. To minimise edge effects caused by atmospheric air incursion at the sides, the blocks fitted tightly into the boxes. Two plastic collars 10 cm in diameter were installed in each block of soil, a ’shallow’ one to a depth of 2.5 cm and a ’deep’ one to 7.5 cm. The two collars were placed symmetrically to be as far from the edges, and each other, as possible. Vegetation (which was almost entirely perennial grass Anthoxanthum odoratum) inside the collars was removed by clipping at soil level with scissors. The mesocosms were housed in a climate-controlled growth room with a 12 h photoperiod and constant temperature of 20 °C. The soil moisture content was maintained at approximately 20–30 % by watering every few days when the soil efflux was not being sampled. After one month of acclimatisation, the soil-surface CO2 efflux was sampled using open dynamic chambers in which the [CO2] was controlled at ambient levels by the automated regulation of a flow of CO2-free air into the chamber and sample air out.[28] Following a stabilisation period of at least 1 h, all the CO2 in the chamber was derived from soil-surface efflux and the chamber flows and [CO2] were in a steady state. Our sampling system allowed four chambers to be deployed at any one time.[12] The efflux was sampled on both the collars in two soil blocks; δ13C of the soil-respired CO2 was measured continuously for 48 h. This was repeated four times over 16 days so that all 16 collars were sampled. Subsequently, the shallow collars were removed from the soil and the chambers placed directly onto the soil surface in the same location for an additional set of eight measurements (one from each of the eight blocks) over 2 days. The isotopic composition of the CO2 was measured immediately by wavelength-scanned, cavity ringdown spectroscopy (CRDS) (Picarro G1101-i; Picarro Inc., Santa Clara, CA, USA) via a gas manifold and data logger that
Root and soil incubations, and tests for air ingress to the chambers, were carried out at the James Hutton Institute experimental site on the Ballogie Estate in Scotland (2° 44’W, 57° 02’N; UK grid reference NO556936). This site is typical of large areas of Scotland and other northern temperate regions. It is entirely dominated by C3 vegetation, has highly organic soil, turnover of organic matter is slow and CO2 efflux rates are low. The site is on a gently sloping, north-facing hillside at an altitude of approximately 230 m; natural vegetation cover is almost entirely heather (Calluna vulgaris); the soil is a humus-iron podzol with a surface organic layer typically 5–10 cm deep. Samples were taken from inside 25 × 70 m fenced plots that contain 6-year-old birch and Scots pine plantations within a larger area of natural, heather-dominated moorland.
H. S. K. Snell, D. Robinson and A. J. Midwood
2344
We wanted to characterise this change in detail, to assess if samples taken soon after soil excavation were representative of in situ microbial respiration. In August 2012, December 2012 and April 2013 approximately 500 g of soil from eight (August and December) or seven (April) locations was sampled and incubated to measure δ13C of the CO2 respired from root-free soil. Soil was excavated and roots and stones quickly removed by hand before the soil was placed in a gastight Tedlar® bag. The bag was evacuated and flushed with CO2-free air repeatedly, about five times in 1 min, until the [CO2] in the bag was 25 μmol mol–1 or less, as measured by syringe sampling an aliquot of gas via a kynar valve in the bag and analysing it with a portable infra-red gas analyser (EGM-4; PP Systems, Amesbury, MA, USA). From the start of the excavation process to this point took about 12 min. After a few minutes of incubation when the [CO2] in the bag became sufficient for isotopic analysis (around 500 μmol mol–1) the air in the bag was transferred to a sealed Tedlar® bag or preevacuated vial (capped with heat treated septa[27]) and analysed by either CRDS (August 2012 and April 2013) or IRMS (December 2012). The bag of soil was then again repeatedly flushed with CO2-free air until the [CO2] was as low as possible (
Revised: 11 August 2014
Accepted: 12 August 2014
Published online in Wiley Online Library
Rapid Commun. Mass Spectrom. 2014, 28, 2341–2351 (wileyonlinelibrary.com) DOI: 10.1002/rcm.7017
Minimising methodological biases to improve the accuracy of partitioning soil respiration using natural abundance 13C Helen S. K. Snell1,2*, David Robinson1 and Andrew J. Midwood2 1
Institute of Biological and Environmental Sciences, University of Aberdeen, Cruickshank Building, St Machar Drive, Aberdeen AB24 3UL, UK 2 Environmental and Biochemical Sciences, James Hutton Institute, Craigiebuckler, Aberdeen AB15 8QH, UK RATIONALE: Microbial degradation of soil organic matter (heterotrophic respiration) is a key determinant of net ecosystem exchange of carbon, but it is difficult to measure because the CO2 efflux from the soil surface is derived not only from heterotrophic respiration, but also from plant root and rhizosphere respiration (autotrophic). Partitioning total CO2 efflux can be achieved using the different natural abundance stable isotope ratios (δ13C) of root and soil CO2. Successful partitioning requires very accurate measurements of total soil efflux δ13CO2 and the δ13CO2 of the autotrophic and heterotrophic sources, which typically differ by just 2–8 ‰. METHODS: In Scottish moorland and grass mesocosm studies we systematically tested some of the most commonly used techniques in order to identify and minimise methodological errors. Typical partitioning methods are to sample the total soil-surface CO2 efflux using a chamber, then to sample CO2 from incubated soil-free roots and root-free soil. We investigated the effect of collar depth on chamber measurements of surface efflux δ13CO2 and the effect of incubation time on estimates of end-member δ13CO2. RESULTS: (1) a 5 cm increase in collar depth affects the measurement of surface efflux δ13CO2 by –1.5 ‰ and there are fundamental inconsistencies between modelled and measured biases; (2) the heterotrophic δ13CO2 changes by up to 4 ‰ within minutes of sampling; we recommend using regression to estimate the in situ δ13CO2 values; (3) autotrophic δ13CO2 measurements are reliable if root CO2 is sampled within an hour of excavation; (4) correction factors should be used to account for instrument drift of up to 3 ‰ and concentration-dependent non-linearity of CRDS (cavity ringdown spectroscopy) analysis. CONCLUSIONS: Methodological biases can lead to large inaccuracies in partitioning estimates. The utility of stable isotope partitioning of soil CO2 efflux will be enhanced by consensus on the optimum measurement protocols and by minimising disturbance, particularly during chamber measurements. Copyright © 2014 John Wiley & Sons, Ltd.
Quantifying CO2 emissions from soil organic matter (SOM) is essential for estimating a landscape’s carbon balance and for assessing the effects of climate, land-use and biodiversity on soil carbon stocks. Soil is the largest terrestrial store of carbon. The flux of CO2 from soils to the atmosphere is estimated at around 98 Pg of carbon per year,[1] the same order of magnitude as the flux attributable to primary productivity and about 10 times the amount that is due to anthropogenic emissions.[2] Carbon is lost from soil via the microbial respiration of SOM (heterotrophic) and added via the addition of plant biomass. Some of the CO2 assimilated photosynthetically by plants is released as CO2 during root respiration (autotrophic) and diffuses out of soil alongside the CO2 produced from SOM. Accurate assessment of the net ecosystem exchange of carbon (NEE, i.e. whether a landscape is a source or sink for CO2) relies on partitioning these autotrophic (RA) and heterotrophic (RH) contributions to total soil CO2 efflux (RS).
Rapid Commun. Mass Spectrom. 2014, 28, 2341–2351
Copyright © 2014 John Wiley & Sons, Ltd.
2341
* Correspondence to: H. S. K. Snell, Institute of Biological and Environmental Sciences, University of Aberdeen, Cruickshank Building, St Machar Drive, Aberdeen AB24 3UL, UK. E-mail: [email protected]
Methods to partition soil respiration often physically remove the autotrophic component in an experimental plot and infer the autotrophic contribution by comparison with an unaltered control plot. Such field techniques have been reviewed in detail elsewhere[3–6] and include root removal, trenching (to sever existing roots and prevent ingrowth of new ones) and stem girdling (whereby a ring of phloem is removed in order to halt the supply of photosynthate to the roots). However, SOM and root respiration are intrinsically linked by the priming effect, whereby the rate of microbial respiration of SOM (RH) is modified by the presence of live roots, which affect the soil environment via their mycorrhizal associations, nutrient uptake, exudation and physical presence.[7] Plant roots also alter the bacterial community structure,[8] soil carbon pools,[6] and moisture and temperature regimes of soil,[9] all of which can influence microbial respiration.[8,10,11] A potentially important limitation of field techniques that remove or sever roots (or reduce root respiration) is that they will also remove this primed component of heterotrophic respiration. This would result in an estimate of heterotrophic respiration that may not be an accurate reflection of the undisturbed flux. Natural abundance 13C offers an alternative means of partitioning the total soil-surface CO2 efflux (RS) and has the advantage that it requires no prior disturbance or alteration
H. S. K. Snell, D. Robinson and A. J. Midwood of the site. Briefly, this method relies on measuring the 13C content of RS and then of the CO2 respired from roots and root-free soil. A simple linear mixing model is then used to apportion the total flux into its RA and RH components. RS is typically sampled by placing a chamber on the soil surface. Chambers may be closed or open in design and sampled either discretely or continuously. Open, dynamic chambers, in which the headspace is supplied with CO2-free air (so that after a period of equilibration or flushing the CO2 in the chamber is derived only from soil-surface efflux and does not contain any atmospheric CO2), are useful as they minimise the error in measurement of δ13CRS.[12] Following sampling of the surface efflux, separated roots and root-free soil are incubated in vials or flasks,[9,13–15] or in gastight bags,[16–19] from which atmospheric CO2 is removed so that CO2 derived entirely from the root or soil respiration can accumulate. These three CO2 samples are analysed for their 13C/12C composition (δ13C) and the heterotrophic flux (RH) is calculated using the following equations: f RH ¼
δ13 CRS δ13 CRA δ13 CRH δ13 CRA
and f RA ¼ 1 f RH
2342
where fRH and fRA are, respectively, the fractions of total soil-surface efflux attributable to heterotrophic and autotrophic respiration; δ13CRS, δ13CRA and δ13CRH are the isotopic signatures of the total soil-surface efflux, autotrophic (soil-free root) respiration and heterotrophic (root-free soil) respiration. Heterotrophic flux is the product of fRH and total soil-surface flux, RS. Most natural and semi-natural temperate ecosystems are dominated by plants that use C3 photosynthesis and consequently the accumulated SOM is derived entirely from C3 vegetation. Microbial processing of dead plant material leads to a slight 13C-enrichment of the remaining SOM and as a result the difference between the isotope ratios of RA and RH is typically 2–8 ‰.[9,16,17,20] This range, although typical for natural abundance carbon isotopes, is small. Therefore, the integrity of isotopic partitioning of the CO2 flux depends heavily on the accuracy and precision of the δ13CO2 measurements of RS, RA and RH.[21] In some agricultural and naturally transitional ecosystems, C4 plants occur on historically C3 soil. Just as when plants are artificially labelled with 13C-depleted CO2, the difference between the RH and RA end-member δ13CO2 is greater in these cases and stable isotope partitioning estimates will probably be more accurate. However, these systems are uncommon, particularly in natural, temperate landscapes. Stable isotope partitioning studies report very varied endmember δ13CO2 values and source contributions. While some of this variability undoubtedly reflects the natural differences between ecosystems, some of it may well be due to methodological differences between studies. Coupled with the fact that stable isotope partitioning is not always successful (i.e. if the surface efflux δ13CO2 falls outside the range of the end members),[16] there is clearly scope to improve the utility of this approach by refining the methodology. Methodological biases can affect all three of
wileyonlinelibrary.com/journal/rcm
the δ13CO2 measurements required for stable isotope partitioning and are mainly associated with differences in chamber design and collar usage, and with disturbance effects during the separation of roots and soil. Chamber measurements of soil-surface efflux δ13CO2 are not straightforward due to the way in which the different isotopologues of CO2 diffuse through the soil. While the δ13C of the soil-surface CO2 efflux is, of course, defined by that of the combined CO2 sources, the different molecular weights of 12CO2 and 13CO2 mean that diffusive fractionation causes the CO2 within the soil profile to be 13C-enriched by around 4.4 ‰ compared with that of the sources and efflux (this fractionation is in fact a function of the δ13C of the respired CO2 and the soil [CO2], and varies by around 0.5 ‰ from this value). Further 13C-enrichment occurs in the upper soil layers due to mixing with (–8 ‰) atmospheric air. The 13C-enrichment in the soil profile is thus 4.4 ‰ at depth and is greater (approaching an absolute value of -8 ‰) close to the soil surface.[22,23] The very presence of the chamber can disturb the diffusion profile and bias the δ13C value of the CO2 diffusing into the chamber. The δ13CO2 of the measured efflux can be affected by deployment time, non-ambient chamber pressure and chamber [CO2]. These factors vary with chamber design and are the reason why different chamber methodologies record different values of efflux δ13CO2, both theoretically and in practice.[12,24–26] Rigid collars installed in the soil surface are commonly used to site chambers when sampling soil-surface CO2 efflux and are intuitively thought necessary to prevent the lateral ingress of atmospheric CO2 to the chamber. However, these too can influence the measurement of δ13CO2. For example, when CO2 diffusion from soil is modelled, open dynamic chambers tend to equilibrate with δ13C of the CO2 at the base of the collar, rather than with that of the true surface efflux.[25] Although the impact of collar depth on the measured δ13CO2 has been considered theoretically, it has not been tested in practice, is not standardised and is often overlooked when results are reported. Measurements of end-member δ13CO2 are similarly prone to disturbance effects. The process of excavating soil and separating it from roots exposes labile carbon substrates, depleted in 13C, which are preferentially respired by soil microbes. This causes an immediate and progressive decrease in δ13CO2 during incubation.[19] Root respiration exhibits species-specific changes to δ13CO2 following excision.[27] Thus, the method of soil and root separation (sieving or hand sorting) as well as the duration of the incubation will potentially introduce methodological biases to the measured end-member values. Finally, soil respiration studies are increasingly using cavity ringdown spectroscopy (CRDS) for isotopic analysis. While these systems have the advantage of portability and ease of use, the precision and accuracy of the analyses need to be optimised by a combination of operational protocols and post-analysis data handling; an issue that has yet to receive much consideration in the literature. The aim of this study was to improve the accuracy of isotopic flux partitioning by identifying and minimising methodological errors associated with some commonly used techniques. We tested the effect of collar depth on chamber measurements of δ13CO2 using grass mesocosms in the laboratory. We expected that the chamber measurement
Copyright © 2014 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2014, 28, 2341–2351
Methods in partitioning soil respiration using natural abundance13C would reflect δ13CO2 at the base of the collar and that, due to mixing with atmospheric air in the upper soil, shallower collars would give more 13C-enriched measurements than deeper ones. Furthermore, we expected that chambers used without collars would accurately measure δ13C of the surface efflux CO2, which is more 13C-depleted than the subsurface CO2 at any depth. We also tested whether chambers used without collars in the field were subject to contamination with atmospheric air. We investigated measurement errors during end-member incubations in the field to establish whether CO2 samples taken soon after root and soil excavation were representative of the in situ δ13CO2. We devised and applied correction factors to CRDS analysis to account for analyser drift and non-linearity.
controls the [CO2] in the chambers as well as the air flow to the CRDS analyser; chambers were sampled sequentially for 10 min each, which is enough time for the CRDS to stabilise and give a few minutes (13 analyses) of continuous data at each sampling time. A cylinder of reference gas (530 μmol mol–1 CO2 in a balance of air)[29] was analysed by CRDS every day and all data corrected to isotope ratio mass spectrometric (IRMS) analysis of the same gas using a Thermo Delta PlusXP isotope ratio mass spectrometer interfaced to a Gas Bench II (both from Thermo, Bremen, Germany) and a PAL autosampler (CTC Analytics, Zwingen, Switzerland)). Means of between 351 and 559 (in one case 273) daytime measurements of the δ13CO2 from each location were compared with Anova in Minitab 16 (Minitab Inc., Coventry, UK).
EXPERIMENTAL
Field site
Collar depth
Rapid Commun. Mass Spectrom. 2014, 28, 2341–2351
Air ingress On five days between the 25th and 30th April 2013 we field tested whether chambers used without collars were subject to lateral air ingress causing a difference in δ13CO2. We modified two of four chambers by the addition of a 4 cm wide foot on the chamber edge. The underside of the foot was made of flexible closed cell foam; when the chamber was weighted, the foot made a seal across its whole surface, even on uneven ground. The diffusive path length through the soil, from the atmosphere outside to the headspace inside the chamber, was thus 4 cm, the same as would be created by using a collar 2 cm deep but without severing roots or penetrating the soil. We compared these modified chambers with unmodified chambers placed directly on the soil surface without collars. Wet sand was placed around the chamber edge or foot and a 3 kg weight was placed on top of the chamber to ensure a good seal with the ground. Sixteen measurements of surface efflux were made both with and without the modified chamber foot. A general linear model with time of measurement as a random factor was used to test for a difference in δ13CO2 values between the modified and unmodified chamber measurements. Heterotrophic end member The heterotrophically respired δ13CO2 changes rapidly during incubation because soil disturbance exposes labile, 13 C-depleted substrates that are preferentially respired.[19]
Copyright © 2014 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/rcm
2343
To investigate the influence of collar depth on the measured δ13C of soil-respired CO2, eight monoliths of grassland soil (Countesswells association, from the JHI Aberdeen site) measuring 60 × 40 cm and approximately 20 cm deep were extracted intact and placed in plastic boxes of the same dimensions. In our experience this is close to the maximum size of monolith that can be removed intact at this site. To minimise edge effects caused by atmospheric air incursion at the sides, the blocks fitted tightly into the boxes. Two plastic collars 10 cm in diameter were installed in each block of soil, a ’shallow’ one to a depth of 2.5 cm and a ’deep’ one to 7.5 cm. The two collars were placed symmetrically to be as far from the edges, and each other, as possible. Vegetation (which was almost entirely perennial grass Anthoxanthum odoratum) inside the collars was removed by clipping at soil level with scissors. The mesocosms were housed in a climate-controlled growth room with a 12 h photoperiod and constant temperature of 20 °C. The soil moisture content was maintained at approximately 20–30 % by watering every few days when the soil efflux was not being sampled. After one month of acclimatisation, the soil-surface CO2 efflux was sampled using open dynamic chambers in which the [CO2] was controlled at ambient levels by the automated regulation of a flow of CO2-free air into the chamber and sample air out.[28] Following a stabilisation period of at least 1 h, all the CO2 in the chamber was derived from soil-surface efflux and the chamber flows and [CO2] were in a steady state. Our sampling system allowed four chambers to be deployed at any one time.[12] The efflux was sampled on both the collars in two soil blocks; δ13C of the soil-respired CO2 was measured continuously for 48 h. This was repeated four times over 16 days so that all 16 collars were sampled. Subsequently, the shallow collars were removed from the soil and the chambers placed directly onto the soil surface in the same location for an additional set of eight measurements (one from each of the eight blocks) over 2 days. The isotopic composition of the CO2 was measured immediately by wavelength-scanned, cavity ringdown spectroscopy (CRDS) (Picarro G1101-i; Picarro Inc., Santa Clara, CA, USA) via a gas manifold and data logger that
Root and soil incubations, and tests for air ingress to the chambers, were carried out at the James Hutton Institute experimental site on the Ballogie Estate in Scotland (2° 44’W, 57° 02’N; UK grid reference NO556936). This site is typical of large areas of Scotland and other northern temperate regions. It is entirely dominated by C3 vegetation, has highly organic soil, turnover of organic matter is slow and CO2 efflux rates are low. The site is on a gently sloping, north-facing hillside at an altitude of approximately 230 m; natural vegetation cover is almost entirely heather (Calluna vulgaris); the soil is a humus-iron podzol with a surface organic layer typically 5–10 cm deep. Samples were taken from inside 25 × 70 m fenced plots that contain 6-year-old birch and Scots pine plantations within a larger area of natural, heather-dominated moorland.
H. S. K. Snell, D. Robinson and A. J. Midwood
2344
We wanted to characterise this change in detail, to assess if samples taken soon after soil excavation were representative of in situ microbial respiration. In August 2012, December 2012 and April 2013 approximately 500 g of soil from eight (August and December) or seven (April) locations was sampled and incubated to measure δ13C of the CO2 respired from root-free soil. Soil was excavated and roots and stones quickly removed by hand before the soil was placed in a gastight Tedlar® bag. The bag was evacuated and flushed with CO2-free air repeatedly, about five times in 1 min, until the [CO2] in the bag was 25 μmol mol–1 or less, as measured by syringe sampling an aliquot of gas via a kynar valve in the bag and analysing it with a portable infra-red gas analyser (EGM-4; PP Systems, Amesbury, MA, USA). From the start of the excavation process to this point took about 12 min. After a few minutes of incubation when the [CO2] in the bag became sufficient for isotopic analysis (around 500 μmol mol–1) the air in the bag was transferred to a sealed Tedlar® bag or preevacuated vial (capped with heat treated septa[27]) and analysed by either CRDS (August 2012 and April 2013) or IRMS (December 2012). The bag of soil was then again repeatedly flushed with CO2-free air until the [CO2] was as low as possible (
