Printer Friendly

Quantifying the allocation of soil organic carbon to biologically significant fractions.

Received 21 December 2012, accepted 21 July 2013, published online 20 December 2013

Introduction

Soil organic carbon (OC) is composed of a variety of materials that originate predominantly from carbon captured by photosynthesis and subsequently deposited as residues within the soil matrix or on the soil surface. The range of photosynthetic organisms and the mechanisms employed to grow and persist under variable environmental conditions lead to diversity in the biochemicals created and their relative proportions within organic materials received by soils (Kogel-Knabner 2002). This diversity is expanded as soil-based decomposer communities process and transform the incoming materials, synthesise additional biochemicals required for their maintenance and growth, and excrete products of metabolic activity (Grandy and Neff 2008). Associated with the decomposition of incoming organic materials is a diminution of particle size and a reduction in decomposability due to preferential loss of more labile components and potential interactions with soil minerals (e.g. adsorption onto mineral surfaces or occlusion within aggregates) (Golchin et al. 1997; Baldock and Skjemstad 2000; Grandy and Neff 2008).

Simulation models developed to predict the dynamics of soil OC attempt to deal with the diversity of composition and its impact on turnover through the creation of conceptual pools of carbon that cycle at different rates (e.g. Jenkinson 1990; Parton et al. 1998; Petersen et al. 2005). Typically, these models consist of one or more small but rapidly cycling soil microbial biomass pools, one or more pools related to incoming plant residues, and two or more pools of more stable soil OC with all of the component pools being differentiated by their relative rates of decomposition. Jenkinson (1990) introduced an inert organic matter fraction to the Rothamsted soil carbon model (RothC) in order to adequately explain both variations in soil carbon stocks and radiocarbon data collected from the Broadbalk field experiment. Skjemstad et al. (2004) proposed the use of measureable fractions of soil OC to replace the conceptual model pools and demonstrated that the RothC RPM, HUM, and IOM pools could be replaced, respectively, with measured fractions of particulate OC (soil OC associated with particles >53 [micro]m), humus OC (soil OC associated with particles [less than or equal to] 53 [micro]m excluding charcoal carbon), and charcoal carbon (soil OC associated with particles [less than or equal to] 53[micro]m and having a chemical structure consistent with charcoal). An ability to replace the RothC conceptual pools with measureable fractions was also demonstrated by Zimmermann et al. (2007), although the fractions measured differed to those used by Skjemstad et al. (2004). The ability to measure the quantity and model the dynamics of soil OC fractions suggests that they represent organic components of biological significance, at least in terms of their relative rates of decomposition.

In the fractionation scheme proposed by Skjemstad et al. (2004), the content of soil OC and its component fractions were calculated as follows:

* soil OC content was measured on the sample before fractionation using an automated Dumas combustion carbon analyser,

* the particulate OC fraction was measured based on the amount of OC retained on a 53-[micro]m sieve;

* the charcoal C content was measured as the amount of OC in the [less than or equal to] 53 [micro]m fraction found to exist in the form of charcoal as assessed by [sup.13]C nuclear magnetic resonance (NMR) analysis after exposure to a photo-oxidation process; and

* the humus carbon was calculated as the soil OC minus the particulate and charcoal fractions.

The approach of Skjemstad et al. (2004) did not quantify the presence of charcoal carbon in the >53 [micro]m particulate fraction. Although Skjemstad et al. (2004) demonstrated that this was appropriate for the soils used in their study, it should not be ubiquitously assumed, particularly for soils with a history of forest vegetation and fire. Given the method used by Skjemstad et al. (2004) to calculate the content of humus carbon, uncertainty in estimates of humus carbon content equated to the additive uncertainty associated with the measurements made to quantify all other carbon components, and carbon balance calculations always returned a value of 100%. Additionally, the size fractionation step within the methodology of Skjemstad et al. (2004) was labour-intensive and subject to an operator dependence. The objectives of this study were: (i) to assess a new automated protocol for size fractionating soil OC, (ii) to quantify the presence of charcoal-like carbon in both the coarse and fine materials isolated by size fractionation, (iii) to provide a direct measurement of the humus carbon fraction, and (iv) to allow appropriate assessments of the carbon balance associated with the fractionation process. The nomenclature and acronyms used in this study to describe the various forms of soil carbon and its allocation to carbon fractions are provided in Appendix 1.

Materials and methods

Soil samples

Two separate sets of soil samples were used in this study. The preliminary comparison of the manual method of Skjemstad et al. (2004) and new, more automated soil carbon fractionation methodology was completed using large samples (>10kg) collected from five different soil types found in Australia. Sampling depth, Australian Soil Classification (Isbell 2002), and some additional chemical and physical properties of each soil are provided in (Table 1). Each of these samples was airdried, sieved to [less than or equal to] 2 mm, and mixed to produce a homogenous sample. All aliquots of soil used in the fractionation method comparison were drawn from the homogenised soil.

After comparing the fractionation methods, the automated methodology was applied to 312 soil samples collected from agricultural fields by the Australian Soil Carbon Research Program (SCARP: www.csiro.au/science/Soil-Carbon-ResearchProgram). To collect soil samples in the SCARP, agricultural fields to be sampled were identified, a soil sampling site 25 m by 25 m was randomly placed within each field, a grid 5 m by 5 m was delineated within the sampling site, and 10 grid intersection points were randomly selected from the possible 36 points to define soil-sampling locations. At each sampling location, soil was collected from the 0-10, 10-20, and 20-30 cm depth layers using a soil corer with a minimum internal diameter of 40 mm. The 0 cm depth was set at the surface of the mineral soil, so any organic layers or surface residues were excluded. For 95% of the sampling sites, the 10 samples collected from each depth were composited into one bulk sample. At 5% of the sampling sites, soil samples collected from each sampling location were kept separate. All collected soil samples were air-dried, crushed, and sieved to [less than or equal to] 2 mm. All subsequent measurements were performed on aliquots of the [less than or equal to] 2 mm soil split from the main sample using a riffle box (12 by 13 mm slotted box; Civilab Australia, Sydney) to minimise bias when separating subsamples for subsequent analyses.

The 312 soil samples selected for OC fractionation in this study were defined based on measured OC contents. As analyses of the OC content of groups of soil samples collected from different regions of Australian agricultural soils were completed, ~5% were identified for fractionation as follows. The samples were ranked from lowest to highest on the basis of measured OC contents; the range of OC contents was determined; the OC range was divided by the number of samples to be selected minus 1 to define an OC content increment; the samples with OC contents corresponding to the minimum, multiples of the increment up to the maximum, and the maximum were identified and selected for analysis. This process resulted in a set of samples with OC contents spread evenly across the range of OC contents measured for each group of samples. The process was repeated for the groups of samples submitted for analysis within the SCARP program. Although attempts were made to cover the range of soil types, depths, and OC contents exhibited by the SCARP samples, due to the timing of sample collection and the time required to complete the fractionation analyses, the samples included are biased towards the initial groups of soil samples submitted for analysis. Additionally, carbonatecontaining soils were poorly represented.

Analysis of total carbon, organic carbon, and inorganic carbon contents of soil [less than or equal to] 2 mm

All total carbon (TC), OC, and inorganic carbon (IC) contents of the [less than or equal to] 2 mm soils were determined on a 10-g aliquot of air- dried soil split from the main sample with a riffle box. The 10-g aliquots were finely ground using an MM400 Mixer Mill (RETSCH, Haan, Germany) equipped with 35-mL zirconia grinding jars and zirconia yttrium-stabilised grinding balls, set at a mixing oscillation frequency of 28 Hz for 3 min. Prior to conducting any carbon analyses, all ground samples were tested for the presence of IC. Approximately 0.5 g of the ground soil was placed in a ceramic or plastic well and a few drops of 1 M hydrogen chloride (HCl) were placed directly on the sample. Samples showing visible effervescence were recorded as having a positive fizz test.

The TC content of all ground samples was determined using either a LECO CNS2000 for ground samples returning a negative fizz test or a LECO C-144 for ground samples returning a positive fizz test (LECO Corporation, St. Joseph, MI, USA). For both carbon analysers, a set of five standard soils was generated as working standards. All five soils were rigorously tested against certified standards by repeated analysis. Each set of samples analysed on the carbon analysers consisted of ~32 unknown samples, 4 or 5 calibration standard soils at the beginning and end of the run, and reference samples run after every 10 unknown samples. The standard soil used for each run was chosen to best match the expected range in carbon concentrations. Calibration and drift correction, if needed, were performed offline in a spreadsheet. Ten per cent of unknown samples were run in duplicate. The long-term mean relative standard deviation (%RSD) calculated according to Eqn 1 for TC content of the standard soils and the repeat unknown samples was 0.72% and 1.11%, respectively:

%RSD = (standard deviation/mean) x 100 (1)

For ground samples returning a negative fizz test, OC content was equated to TC content and the IC content was assigned a value of 0 mg IC/g soil. Ground samples that returned a positive fizz text were analysed a second time on the C-144 analyser after applying a pre-treatment to remove IC. The IC was removed by placing 0.8 g of ground sample into a nickel-lined ceramic LECO boat. The nickel-lined LECO boats containing ground samples were placed on a hot plate set to 100[degrees]C, and 1 mL of 5-6% [H.sub.2]S[O.sub.3] was added to each sample. When the sampled dried, a further 1 mL of [H.sub.2]S[O.sub.3] was added to the sample and the sample left to dry. Acid additions continued until no effervescence was noted when 1 mL of [H.sub.2]S[O.sub.3] was added after which one additional [H.sub.2]S[O.sub.3] treatment was applied. Once all ground samples in a batch were dry, they were left to cool overnight and analysed on the LECO C-144 with a wad of zinc wool added to the top of the water vapour trap to remove any sulfur that may corrode the system. The mass of carbon detected by the LECO C-144 analyser was used along with the initial weight of ground sample added to the nickel-lined LECO boat to calculate the OC content. The IC content of the ground samples returning a positive fizz test was calculated as the difference between the TC and OC contents. When the difference between TC and OC contents was <1.0 mg C/g soil, the 1C content was considered to be below detection limits and was assigned a value of 0 mg C/g soil.

All TC, OC, and 1C contents were corrected to an oven-dry equivalent mass using the results obtained for the air-dried ground samples and their gravimetric water contents ([[theta].sub.m]). Gravimetric water contents were determined using a 20-g aliquot split from the main sample using a riffle box. The 20-g sample was weighed out into an aluminium foil weighing tray, placed in an oven at 105[degrees]C for 16-24 h, allowed to cool in a desiccator, and then re-weighed. Oven-dry equivalent values of TC, OC, and IC were calculated using Eqn 2, where COD is the oven-dry equivalent content (g C/kg soil), CAD is the air-dry content measured (g C/kg air-dry soil), and [[theta].sub.m] is the measured gravimetric water content of the air-dried soil:

[C.sub.OD] = [C.sub.AD](1 + [[theta].sub.m]) (2)

Fractionation of organic carbon present in soil [less than or equal to] 2 mm Manual fractionation

The manual fractionation process was completed in a manner similar to that used by Skjemstad et al. (2004) (Fig. 1a). Soil samples (10g soil [less than or equal to] 2mm) were added to 40mL of a 50g[L.sup.-1] solution of sodium hexametaphosphate and dispersed by shaking on a flatbed orbital shaker (Promax 1020; Heidolph Instruments GmbH, Schwabach, Germany) overnight ([greater than or equal to]14h) at 180rpm and an amplitude of ~2.5 cm. The samples were passed through a nest of two sieves (200 and 53 [micro]m). Soil material remaining on the sieves was manually moved around with a spatula to ensure that no aggregates remained. All materials remaining on the sieves (coarse fraction >53 [micro]m) were quantitatively washed into pre-weighed 500-mL LDPE (low-density polyethylene) bottles. The materials passing through the sieves (fine fraction [less than or equal to] 53 [micro]m) were also washed quantitatively into separate 500-mL LDPE bottles. All fractions were frozen, lyophilised until completely dry, and weighed. Lyophilised samples were removed from the bottles. The coarse fractions were homogenised and ground to a fine powder in a ring mill (Standard Ring Mill, SRM-RC-3P; Rocklabs Ltd, Auckland, New Zealand) equipped with a tungsten carbide head (TC-40-BLP) for 1 min. The fine fractions were homogenised and ground by hand using a mortar and pestle.

Automated fractionation

In the automated fractionation process Fig. 1b, soil samples were dispersed as described for the manual process but using a 5 g [L.sup.-1] sodium hexametaphosphate solution. After dispersion, the sample was passed through a 50-[micro]m sieve using an automated wet sieving system (Vibratory Sieve Shaker Analysette 3 PRO; FRITSCH GmbH, Idar-Oberstein, Germany) with two water nozzles fixed in the Perspex lid. Water was supplied to the sieve shaker using a peristaltic pump (Masterflex L/S Model 7553-79 attached to Masterttex L/S Modular Controller; Cole-Parmer, Vernon Hills, IL, USA). Operating parameters used for the sieve shaker system were as follows: interval 20 s; minimum sieving time 3 min (3 min or until drainage water becomes clear); amplitude 2.5 mm; water spray rate ~150mL/min. Sieving was deemed complete when the stream of water exiting the sieving system was clear. The coarse sample retained on the 50-[micro]m sieve was visually inspected to ensure that all fine material had passed through the sieve. If not, the system was reset and run again. Soil materials collected on the sieve (coarse fraction >50 [micro]m) and passing through the sieve (fine fraction [less than or equal to] 50 [micro]m) were added to separate 500-mL LDPE bottles, frozen, lyophilised until completely dry, and weighed. Lyophilised coarse and fine fractions were removed from the bottles and then homogenised and ground as described for the manual fractionation.

Organic carbon contents of collected fractions

The OC contents of the coarse and fine soil fractions were measured as described for the ground whole soils [less than or equal to] 2 mm. Any fractions that tested positive to the presence of IC were treated with 5-6% [H.sub.2]S[O.sub.3] (as described for the soils [less than or equal to] 2 mm) to remove carbonate carbon before analysis. Results obtained from the LECO analysers were taken as OC contents, given that all analysed fractions were devoid of 1C when analysed. A correction for gravimetric water content (Eqn 2) was not applied to the measured OC values obtained for the coarse or fine fractions since they were dried completely by lyophilisation before analysis.

Comparison of manual and automated fractionation methods

Two assessments of the implication of automating the size fractionation process were completed. In the first, one operator analysed four replicates of each of the five soils listed in Table 1 using the manual and automated fractionation methods. The contents of coarse OC per unit mass of soil [less than or equal to] 2mm were calculated according to Eqn 3 using the OC content of the coarse fraction (g OC/kg coarse fraction) and the proportional mass ([MF.sub.2000]) of the coarse fraction (g coarse fraction/g [less than or equal to] 2 mm soil).

Coarse OC (g C/kg soil [less than or equal to] 2 mm) (3) = OC content coarse fraction x [MF.sub.2000]

Results obtained for the two fractionation methodologies were analysed according to a 5 (soil types) x 2 (methods of sieving) factorial design. To quantify the extent of operator dependence on the results obtained from each fractionation method, three different operators analysed four replicates of two soils using both sieving methodologies and the data were analysed as a 3 (operators) x 2 (soil types) x 2 (methods of sieving) factorial design. All statistical analyses were completed using Statistica 8 (Statsoft Inc. 2007):

Allocation of organic carbon in the coarse and fine fractions to resistant organic carbon

The proportion of resistant OC (ROC) within the OC present in the coarse and fine fractions separated for 312 soils was determined using solid-state [sup.13]C NMR. The ROC fraction was measured as the amount of poly-aryl carbon existing in addition to any aryl C that could be defined as lignin. It is likely that the ROC was dominated by charcoal and charred plant residues; however, the potential presence of poly-aryl structures that may have been created in response to chemical reactions in the soil, e.g. various abiotic condensation theories described by Tan (2003), cannot be ruled out as an additional contributor to the ROC fraction.

A mass of coarse- or fine-fraction OC large enough to allow [sup.13]C NMR analysis to be completed was obtained by fractionating several 10-g aliquots of the unground soil [less than or equal to] 2mm. One to 10 aliquots of unground soil [less than or equal to] 2mm were dispersed, sieved, and combined into an accumulated coarse or fine fraction. The OC present in the accumulated coarse fractions was separated from most of the mineral grains using a stream of water to push the organic materials and some fine sand out of the coarse sand particles and into a separate container. Separation of coarse OC from sand particles was required to concentrate the carbon to a point where [sup.13]C NMR spectra could be acquired. Care was taken in the separation process to ensure that all OC particles were transferred. The resulting sample was frozen, lyophilised, and ground in preparation for [sup.13]C NMR analyses. The accumulated fine fractions were frozen, lyophilised, and treated with 2% hydrofluoric acid (HF) according to the methodology presented by Skjemstad et al. (1994) to remove paramagnetic materials and concentrate the OC for subsequent [sup.13]C NMR analyses.

The coarse and fine fractions separated from all 312 soils were analysed by [sup.13]C NMR, resulting in the acquisition of 624 NMR spectra. All [sup.13]C NMR spectra were acquired on a 200 Avance spectrometer (Bruker Corporation, Billerica, MA, USA) equipped with a 4.7 T, wide-bore superconducting magnet operating at a resonance frequency of 50.33 MHz. Weighed samples (100-600mg) with known carbon contents were packed into 7-mm-diameter zirconia rotors with Kel-F[R] end caps and spun at 5 kHz. Kel-F inserts were used to fill any gaps and place the sample in the middle of the rotor when samples did not fill the rotor completely. Chemical shift values were calibrated to the methyl resonance of hexamethylbenzene at 17.36ppm, and a Lorentzian line broadening of 50Hz was applied to all spectra.

Four separate [sup.13]C NMR experiments were performed. A cross-polarisation [sup.13]C NMR (CP) analysis using a pulse of 3.2[micro]s, 195W, and 90[degrees], a contact time of 1 ms, and a recycle delay of 1 s was applied to all fractions. Scans (10 000-30 000) were collected for the CP analyses. The number of scans was increased as the amount of carbon contained in the rotor declined. The adequacy of the 1-s delay was confirmed by determining the value of [T.sub.1]H for ~50% of the samples using an array of inversion recovery times (0.001, 0.01, 0.05, 0.1, 0.2, 0.3, 0.4 s) and a recycle delay of 5 s. Calculated [T.sub.1]H values were 5-10 times shorter than the 1-s recycle delay used, indicating that saturation did not affect the acquired CP signal intensities. A variable spin-lock experiment using an array of spin-lock times (1, 2, 6, 10, 15, 20ms), a contact time of 1 s, and a recycle delay of 1 s was applied to -75% of the samples to calculate sample-specific values of [T.sub.1][rho]H. A direct polarisation [sup.13]C NMR analysis (DP) was completed on a subset of 38 fractions selected to be indicative of the range of aryl C contents obtained from the CP analyses. For the DP analyses, a pulse of 4 [micro]s, 180 W, and 90[degrees] and a recycle delay of 90 s were used and 1000 transients were collected.

All spectral processing was completed using the Bruker TopSpin 3 software. Absolute signal intensities were acquired for all samples and an empty rotor. Sample intensities were corrected for variations in detection efficiency due to the use of inserts and divided by the number of transients collected. Empty rotor background signals were subtracted from the spectra obtained for all samples, and the resultant spectra were integrated using the chemical shift limits defined in Table 2. Signal intensity found in spinning side bands was allocated back to their parent resonances according to the calculations presented by Baldock and Smernik (2002). The amounts of [sup.13]C NMR signal intensity in the aryl and O-aryl regions of the acquired spectra were used to estimate the proportion of OC present in the fractions that should be allocated to the ROC fraction. The process developed to allocate OC to ROC is presented and discussed in subsequent sections of this paper.

The proportion of OC found in each sample that was observable in the CP and DP NMR analyses ([[phi].sub.CP] and [[phi].sub.DP], respectively) was calculated using the methodology of Smernik and Oades (2000b, 2000a). Glycine was used as an external intensity standard against which the observability of sample OC was compared. Where [T.sub.i][rho]H values were not directly measured for a sample, the average value of 4.5 ms obtained for samples analysed in this study was used.

Fractionation calculations

In the fractionation methodology developed in this study, the contents of particulate OC (POC), humus OC (HOC), and ROC, expressed in g C/kg soil, were calculated according to Eqns 4-6:

POC = (2000 - 50 [micro]m OC)(1 - [f.sub.ROC2000])([MF.sub.2000]) (4)

HOC = ([less than or equal to] 50[micro]m OC)(1 -[f.sub.ROC50])([MF.sub.50]) (5)

ROC = [ROC.sub.2000] + [ROC.sub.50] = (2000 - 50[micro]m OC)([f.sub.ROC2000])([MF.sub.2000]) (6)

+ ([less than or equal to] 50 [micro]m OC)([f.sub.ROC50])([MF.sub.50])

where 2000-50 [micro]m OC is the measured OC content in the coarse 2000-50 [micro]m material (g OC/kg 2000-50 [micro]m fraction); [f.sub.ROC2000] is the fraction of coarse 2000-50 [micro]m OC found to be ROC as determined by [sup.13]C NMR (g [ROC.sub.2000]/g 2000-50 [micro]m OC); [MF.sub.2000] is the fraction of total soil mass found in the coarse 2000-50[micro]m fraction (g 2000-50 [micro]m fraction/g [less than or equal to]2 mm soil); [less than or equal to]50 [micro]m OC is the measured OC content in the fine [less than or equal to]50 [micro]m material (g OC/kg [less than or equal to] 50 [micro]m fraction); [f.sub.ROC50] is the fraction of fine [less than or equal to]50 [micro]m OC found to be ROC as determined by [sup.13]C NMR (g [ROC.sub.50]/g [less than or equal to]50 [micro]m OC); and [MF.sub.50] is the fraction of total soil mass found in the fine [less than or equal to]50 [micro]m fraction (g [less than or equal to]50 [micro]m fraction/g [less than or equal to] 2 mm soil).

The methodology developed and used in this study (Fig. l b) therefore differs from that previously established by Skjemstad et al. (2004) in several aspects: ROC present in the coarse fraction was accounted for; the content of ROC was calculated as the sum of that present in both the coarse and fine factions; the content of HOC was derived from measured values and not calculated as a difference between OC, POC, and ROC contents; and an estimate of the recovery of the [less than or equal to]2 mm OC within the fractions could be defined.

Results and discussion

Comparison of manual and automated size fractionation

The influence of the size fractionation method used (manual v. automated) was assessed by quantifying the amount of coarse OC retained on the 50-[micro]m sieve. The data acquired by one operator across all five soils examined (Table 1) indicated the presence of a significant interaction between soil type and fractionation method (Fig. 2). Importantly, greater amounts of coarse OC were obtained in the automated method than the manual method for all soils. However, the magnitude of the difference between the two methods varied between the soil types. Soil SS7 showed a large difference whereas little difference was evident for the Weisenboden soil. The reduced content of coarse OC using the manual method was attributed to a forcing of organic matter fragments >50 [micro]m in size through the 50-[micro]m sieve as the soil particles on the sieve were moved around with a spatula. The allocation of SOC to the coarse fraction using the automated method was considered more representative of the actual size distribution of the [less than or equal to]2mm SOC. The coefficient of variation for all soils, except the Weisenboden, was lower for the automated than for the manual fractionation method. On average, the automated method produced a coefficient of variation 2.8 times lower than that obtained using the manual method.

The observation that more of the [less than or equal to]2 mm soil OC was allocated to the coarse fraction using the automated method persisted when three different operators analysed two soils using both fractionation methods (Fig. 3). Variation both within and between the operators was 2-3 times lower for the automated method than for the manual method of sieving based on reductions in standard deviations and coefficients of variation. Additionally, the automated method was found to be at least two times faster, and less physically demanding, than the manual method. Based on the improved efficiency, ease of use, and improved intra- and inter-sample precision, the automated method of fractionating soils was adopted and used to fractionate the 312 soils included in this study. The automation of the fractionation process would also facilitate the generation of consistent results across multiple laboratories if the fractionation process were to become a routine analysis.

Quantification of ROC content

The ROC fraction in Australian soils is considered to be dominated by poly-aryl structures consistent with charcoal (Skjemstad et al. 1996, 1999, 2004). The NMR signals associated with charcoal are observed in the aryl and O-aryl spectral regions. Due to the low extent of protonation associated with poly-aryl forms of OC such as charcoal, only a fraction (~30%) of poly-aryl OC is detected by CP (Skjemstad et al. 1999; Baldock and Smemik 2002; Smernik et al. 2002). To obtain quantitative detection of poly-aryl forms of OC requires the use of DP; however, due to its lower sensitivity and much longer recycle delays, the use of DP for routine characterisation of soil OC is not possible. Examples of the difference in the detection efficiency of aryl OC by [sup.13]C NMR using the CP and DP analyses are provided in Fig. 4 using the spectra collected for eight fine fractions. Based on these spectra, the amount of aryl OC detected was underestimated by the CP [sup.13]C NMR analysis, leading to lower [[phi].sub.CP] values than the corresponding [[phi].sub.CP] values. Given the higher observability of OC in the DP spectra and the potential for selective loss of signal intensity for particular types of carbon in the CP analyses, the DP spectra were considered more indicative of the actual distribution of the different forms of carbon.

The approach taken to determine the allocation of soil OC to ROC based on the acquired [sup.13]C NMR spectra required a sequential solution to two requirements: (1) development of a procedure to estimate the proportion of CP-detectable OC allocated to poly-aryl ROC; and (2) definition of a correction factor that could be applied to account for the fraction of ROC that was not detected by CP.

Skjemstad et al. (1999) suggested that aryl C signals observed in CP spectra of soil OC originated from two sources: lignin and ROC (poly-aryl char-like materials). They also assumed that the contribution of ROC to the O-aryl spectral region was minimal and that 37% of the aryl C in lignin-type structures was substituted with oxygen and appeared in the Oaryl region. On this basis, Skjemstad et al. (1999) proposed that the contribution of lignin C could be removed from the aryl region by subtracting 1.7 times the O-aryl C, with the remaining aryl C then being ascribed to ROC. A correction factor consistent with only 28% of the ROC being observable in CP relative to DP NMR analyses was then applied by Skjemstad et al. (1999) to derive an estimate of the actual amount of ROC present in samples based on CP spectra. Applying this approach to the 624 CP spectra acquired in this project overcorrected the aryl-C region, resulting in many low and negative allocations of soil OC to the ROC fraction. Further examination revealed that this overcorrection arose due to the assumption that the ROC component did not contribute signal intensity to the O-aryl region. By integrating CP spectra collected for a range of charred materials, Baldock et al. (2004) found that ~15% of the charred C was observed in the O-aryl region.

In this study, it was assumed that lignin and ROC accounted for the majority of signal intensity found in the aryl and O-aryl [sup.13]C NMR regions and that both forms of OC contributed signal to both regions based on a defined average chemical structure. A two-component model (ROC and lignin) was constructed to estimate the allocation of OC to the ROC fraction. The proportional allocations of lignin and ROC to the aryl and Oaryl spectral regions used in this model are provided in Table 3. By setting the proportion of lignin C present in a sample to x and the proportion of ROC to y, the proportions of aryl and O-aryl C in a sample were represented by Eqns 7 and 8, respectively:

Proportion of aryl C = 0.32x + 0.72y (7)

Proportion of O-aryl C = 0.18x + 0.15y (8)

Solving Eqn 8 for the proportion of OC contributed by lignin (x) gave:

x = (O-aryl C 0.15y)/0.18 (9)

= 5.59(O-aryl C) - 0.85y

Substituting Eqn 9 into Eqn 7 gave:

Aryl C = 0.32(5.59(O-aryl C) - 0.85y) + 0.72y = 1.77(O-aryl C) + 0.45y (10)

Rearranging Eqn 10 to solve for y (the proportion of ROC observed in a spectrum) gave:

y = (aryl C - 1.77(O-aryl C))/0.45 (11)

Using Eqn 11 with the measured proportional allocations of NMR signal intensity to the aryl and O-aryl chemical shift regions, estimates of the contribution of ROC to the acquired [sup.13]C NMR spectra were derived. Equation 11 therefore provided a solution to the first requirement for the prediction of ROC from [sup.13]C NMR data.

The second requirement was to derive a correction factor that accounted for the low detection of poly-aryl char-like ROC materials in CP analyses. Provided the NMR instrument and DP pulse sequence are set up correctly, DP analyses should produce estimates of the true allocation of sample OC to the poly-aryl char-like ROC fraction in the absence of paramagnetic metal cations (e.g. [Fe.sup.3+] and [Mn.sup.4+] ). Theoretically, for a CP analysis, the relationship between the measured allocation of sample OC to the ROC fraction is calculated according to Eqn 11. The fraction of ROC observable in a CP analysis ([f.sub.ROC]CP, which equates toy in Eqn 11) and the true allocation ([f.sub.ROC]True) will be defined by the CP observability ([[phi].sub.CP]) of the ROC in a sample according to Eqn 12, where [[phi].sub.CP] can range from 0 to 1. The theoretical relationship between [f.sub.ROC]CP and [f.sub.ROC]True at different values of [[phi].sub.CP] is presented in Fig. 5:

[f.sub.ROC] True = [f.sub.ROC]CP/[[phi].sub.CP]/[f.sub.ROC]CP/[[phi].sub.CP] + 1 (1 - [f.sub.ROC]CP) (12)

The average value of [[phi].sub.CP] for a given set of samples can be derived by running a series of CP and DP analyses on representative samples and defining the optimum [[phi].sub.CP] value that minimises the sum of squares of the differences between the calculated CP corrected values for ROC (estimated values) and corresponding values measured by the DP analysis (best approximation of the true measured values). For this study, the fraction of ROC calculated for the DP spectra was set equal to the true fraction of ROC in Eqn 12. Using the solver in MS Excel, the value of [[phi].sub.CP] was defined by minimising the sum of squares of differences between measured [f.sub.ROC]Tru (where [f.sub.ROC]True = [f.sub.ROC]DP from Eqn 11 calculated for the DP analyses) and the value calculated for [f.sub.ROC]True by substituting the value of [f.sub.ROC]CP (where [f.sub.ROC]CP is calculated from Eqn 11 from a CP analysis) into Eqn 12.

The optimum value obtained for the CP observability ([[phi].sub.CP]) of ROC within the SCARP dataset was 0.51. The data used to derive this value, after optimisation of [[phi].sub.CP], are presented in Fig. 6a. Some of the DP spectra included in the derivation of the CP observability of charcoal C were not optimal, displaying low signal to noise ratios. The optimisation process used to derive a value for [[phi].sub.CP] was repeated using a subset of the 23 best DP spectra and yielded a value for [[phi].sub.CP] of 0.49 (Fig. 6b). Given the close proximity of this value compared with that derived from the entire dataset, the value of 0.51 derived from the entire dataset has been adopted and used in this study. This value of 0.51 for [[phi].sub.CP] obtained for the soils included in this study differed from values of ~0.30 obtained by Smemik and Oades (2000a) and Baldock and Smemik (2002) for pure samples of charcoal.

To calculate the proportion of ROC in the coarse and fine fractions from their respective CP spectra, the following steps were used:

(1) The proportion of total CP signal intensity allocated to the aryl and O-aryl NMR regions was defined.

(2) Eqn 11 was used to calculate the proportion of sample carbon that could be attributed to ROC in the CP spectra ([f.sub.ROC]CP). If the result of Eqn 11 was negative, then the quantity of charcoal C present was assigned a value of 0.

(3) The values calculated for [f.sub.ROC]CP were used along with [[phi].sub.CP] = 0.51 to estimate the true value for the fraction of ROC present in the samples (Eqn 13).

(4) The contents of POC, HOC, and ROC were then calculated by substituting the true values of [f.sub.ROC] obtained for the fractions analysed into Eqns 4, 5, and 6, respectively:

[f.sub.ROC] True = [[f.sub.ROC]CP/0.510]/[[[f.sub.ROC]CP/0.510] + (1 - [f.sub.ROC]CP)] (13)

Presence of ROC within the coarse and fine fractions

Representative CP [sup.13]C NMR spectra acquired for the coarse and fine fractions are presented in Fig. 7a, b. Significant variations were apparent in the magnitude of signal intensity present in the aryl 110 -140ppm chemical shift region. The distribution of [fROC.sub.2000] (the contribution of ROC to the coarse fraction) (Fig. 7c) showed an average value of 0.29 and ranged from 0 to 0.98, indicating that ROC could make substantial contributions to the OC present within the coarse fraction and justifying the more detailed approach taken in the analysis of ROC in this study relative to the previous methodology employed by Skjemstad et al. (2004). For the fine fractions, [fROC.sub.50] averaged 0.33 and ranged from 0.07 to 0.75, again indicating that ROC made substantial contributions to the fine fraction OC.

Allocation of soil OC to POC, HOC, and ROC fractions

Variation in the content of OC within the 312 soils analysed was substantial (1.2-90.9g C/kg soil) and followed a log-normal distribution leading to a positive skew and high kurtosis (Fig. 8). Contents of HOC, POC, and ROC also followed a log-normal distribution, with the maximum contents being >50 times larger than the minimum values (Fig. 9a, b, c). Contents of HOC, POC, and ROC, respectively, varied over ranges of 0.6-39.8 g HOC/ kg soil, <0.1-40.7 g POC/kg soil, and 0.2-20.2 g ROC/kg soil. When expressed as a percentage of soil OC, the distribution of %HOC values moved substantially towards a normal distribution, indicating that OC and HOC were well correlated (r=0.95, P<0.05, Fig. 10a). The HOC accounted for an average of 56.1% of soil OC but ranged from 19.1 to 94.4% with one extreme value of 115% being obtained (not plotted in Fig. 9d). Given the near-normal distribution of the %HOC values, the median value (55.3%) was similar to the mean. The POC ranged from 0.6 to 59.9%, with a mean and median of 19.2 and 17.3%, respectively (Fig. 9e). Although the POC and OC contents were correlated (r=0.87, P<0.05, Fig. 10b), the positive skew associated with the %POC distribution indicated that as the OC content increased, a greater proportional increase in POC content occurred. Thus at the higher OC contents, POC represented a greater fraction of OC than at lower OC contents. The content of the ROC fraction was also correlated with OC content (r=0.90, P< 0.05, Fig. 10c); ROC contributed 6.6-74.2% of soil OC with a mean and median of 26.2 and 23.9% (Fig. 9f), respectively, and a distribution with a skewness intermediate between that obtained for the %HOC and %POC.

Recovery of soil OC within the fractions

The recovery of soil OC within the fractions was calculated as a percentage (Eqn 14) and an absolute deviation (Eqn 15), where POC, HOC, ROC, and OC are all expressed in units of g C/kg soil:

%Recovery = (POC + HOC + ROC)/OC x 100 (14)

Absolute deviation = OC - (POC + HOC + ROC) (15)

When expressed on a percentage basis (Fig. 11a), recoveries of soil OC ranged from 76 to 142% with a mean recovery of 102%, a standard deviation of 7.4%, and 86% of the samples yielding a recovery of 90-110%. Only one value occurred <86% recovery and two values were >116% recovery. Expressing the recovery of soil OC within the fractions as a percentage of the soil OC can be misleading when soil OC contents are low. Under such conditions, small measurement errors or sample variability can translate into a large deviation from 100% recovery. To further assess the recovery of soil OC, the absolute deviation between the sum of OC in the fractions and the soil OC in units of g C/kg soil was calculated (Eqn 15). The frequency distribution of the absolute deviation indicated that, for 89% of the fractionated soils, the sum of the carbon recovered in the HOC, POC, and ROC fractions was within [+ or -] 2 g C/kg soil of the OC content (Fig. 11b). Plotting the magnitude of the absolute deviation as a function of the soil OC content (Fig. 11c) revealed a progressive increase in the magnitude of the absolute deviation with increasing soil OC. Dividing the magnitude of the absolute deviation by soil OC content to produce an estimate of the relative deviation and plotting this value against the soil OC content (Fig. 11d) showed that 84% of the absolute deviations fell within [+ or -] 10% of the amount of OC contained in the soil.

Biological significance of the fractions isolated

The protocol used to quantify the contents of POC, HOC, and ROC in this study was based solely on the physical and chemical characteristics of OC present in a soil. The usefulness of fractionation protocols based on such traits to enhance our understanding of biologically mediated decomposition processes will depend on how successful the protocol has been at isolating OC components with different liabilities or extents of decomposition. The approach used to quantify the amount of ROC present defines the proportion of soil OC found in poly-aryl chemical structures. Poly-aryl structures are consistent with charred plant materials and other forms of black carbon that are considered to be recalcitrant in soil environments and exhibit mean turnover times of centuries to millennia (Skjemstad et al. 1999; Schmidt and Noack 2000; Forbes et al. 2006; Preston and Schmidt 2006). As noted in this study, the fact that ROC often represents a disproportionate fraction of soil OC given the size of ROC input suggests an enhanced resistance to degradation relative to other forms of soil OC. Although some studies have shown an enhanced degradation of components of newly synthesised biochar (Hamer et al. 2004; Marschner et al. 2008; Zimmermann et al. 2012), the use of the solid-state [sup.13]C NMR technique to quantify the proportion of poly-aryl C present in the soil coarse and fine fractions appears consistent with the identification of a biologically resistant form of C. The successful use of the ROC fraction as a replacement for the IOM fraction in the RothC soil carbon cycling model by Skjemstad et al. (2004) provides additional support to the notion that the ROC fraction identified on the basis of a poly-aryl chemical structure has biological significance.

A 50-[micro]m sieve was used to separate non-ROC soil OC into POC and HOC in this study. The use of size fractionation was based on the premise that as pieces of plant residues decompose, a diminution of particle size occurs along with a transfer of degradation products into particles [less than or equal to] 50 [micro]m. Grandy and Neff (2008) developed a model of organic matter decomposition based on work presented by Gregorich et al. (1996) and Denef et al. (2001, 2004) in which the decomposition of POC >53 [micro]m resulted in the production of OC <53 [micro]m. Baldock et al. (1997) showed that changes in [sup.13]C NMR spectra with increasing extent of soil OC decomposition involved a loss of O-alkyl C (indicative of carbohydrates) and an accumulation of alkyl C (indicative of lipids) and that the ratio alkyl C/O-alkyl C could be used to provide an index of the extent of decomposition of soil OC. The exclusion of aryl C and O-aryl C from this ratio means that it is independent of the presence of ROC (poly-aryl C) and influenced only to a minor extent by the presence of lignin. As a result, the alkyl C/O-alkyl C ratios calculated from the [sup.13]C NMR spectra acquired for the coarse and fine soil OC fractions, respectively, were used to estimate the extent of decomposition of the POC and HOC fractions of soil OC. Although there was some overlap of frequency distributions of the alkyl/O-alkyl ratios calculated for the coarse and fine fractions (Fig. 12), 88% of the coarse fractions had values <0.5 and 94% of the fine fractions had values >0.5. This result indicated that the coarse fraction was dominated by less-decomposed, carbohydrate-rich OC and that the fine fraction exhibited a relative depletion of carbohydrate carbon and a greater extent of decomposition. On this basis, passing dispersed soil through a 50-[micro]m sieve was able to separate a more decomposable coarse fraction from a less decomposable fine fraction, providing evidence that the non-ROC soil OC associated with the coarse and fine fractions (the POC and HOC, respectively) was representative of biologically different materials. Further support for the biological differentiation of the POC and HOC fractions was provided by Skjemstad et al. (2004), who obtained different decomposition rate constants for these materials when they were substituted into the RothC soil carbon model calibrated against measured temporal variations.

Summary

Automation of the size fractionation process using a wet-sieving apparatus successfully reduced the coefficient of variance between operators and between replicate analyses conducted by a single operator compared with the previous manual process developed by Skjemstad et al. (2004). Minimising these two sources of variation will be important to ensure consistency across multiple analytical laboratories and to support soil carbon analyses within any future carbon accounting/trading systems that may emerge.

Allocating soil OC to the ROC fraction has proven to be a difficult task, and debate remains as to the appropriateness of the various methodologies (Hammes et al. 2007). In this study, a methodology based on quantifying the presence of poly-aryl carbon identified by solid-state [sup.13]C NMR was used. Contrary to previous work (Skjemstad et al. 2004) where ROC was only found in the particle size fraction [less than or equal to]53 [micro]m, significant quantities of ROC were found in both the coarse (>50 [micro]m) and fine ([less than or equal to]50 p[micro] soil fractions. To ensure adequate allocation of soil OC to the ROC fraction, [sup.13]C NMR spectra had to be collected for both the coarse and fine fractions. The proportion of ROC present in the coarse and fine fractions was estimated by solving a two-component model and then correcting for the average reduction in poly-aryl carbon observability in the typical cross-polarisation [sup.13]C NMR analysis used to quantify the composition of soil OC. Subtracting the [ROC.sub.2000] and [ROC.sub.50] from the OC present in the coarse and fine size fractions, respectively, allowed the contents of POC and HOC to be calculated. Adding the [ROC.sub.2000] and [ROC.sub.50] together provided a value for the total amount of ROC present in a sample. Conducting the fractionation analysis as defined in this study has allowed independent allocations to the HOC, POC, and ROC fractions and carbon balance calculations to be performed. In this study, the sum of the carbon recovered in the HOC, POC, and ROC fractions was within [+ or -] 2 g C/kg soil of the OC content for 89% of the 312 fractionated soils. Significant variations in the allocation of carbon to its component HOC, POC, and ROC fractions were evident across the fractionated soils, and although significant correlation existed between soil OC content and the contents of HOC, POC, and ROC, significant deviations were evident for individual soils, particularly for POC and ROC.

The biological significance of the isolated HOC, POC, and ROC fractions was confirmed by the acquired NMR spectra. The ROC was quantified as the component of the soil OC found in poly-aryl structures consistent with charcoal and char-like structures. The enhanced biological stability of this form of carbon is well documented, although some studies suggest that a small proportion of biochar carbon may be available to decomposition processes. The use of the alkyl C/O-alkyl C ratio demonstrated that, in addition to the physical diminution of particle size, progressing from POC to HOC was associated with an increase in the extent of decomposition. Such changes in extent of decomposition and decomposability across the POC, HOC, and ROC fractions are consistent with the progressive decrease in decomposition rate constants assigned to these fractions when they are incorporated into carbon cycling models.

http://dx.doi.org/10.1071/SR12374

Appendix 1. Nomenclature used to define soil carbon components

Total carbon (TC): the sum of organic and inorganic carbon present in [less than or equal to]2 mm soil.

Organic carbon (OC): the sum of all organic forms of carbon present in [less than or equal to]2 mm soil.

Inorganic carbon (IC): the sum of all inorganic forms of carbon present in [less than or equal to]2 mm soil.

Coarse fraction: all soil materials retained on a 50-[micro]m sieve (2000-50 [micro]m particles) after complete dispersion of [less than or equal to]2 mm soil.

Fine fraction: all soil materials that passed through a 50-[micro]m sieve ([less than or equal to]50 [micro]m particles) after complete dispersion of [less than or equal to]2 mm soil.

Particulate organic carbon (POC): organic carbon associated with coarse fraction (2000-50 [micro]m particles) after the subtraction of organic carbon with a chemical structure consistent with charcoal.

Humus organic carbon (HOC): organic carbon associated with the fine fraction ([less than or equal to]50 [micro]m particles) after subtraction of the organic carbon with a chemical structure consistent with charcoal.

Resistant organic carbon (ROC): organic carbon [less than or equal to]2000 [micro]m found in the coarse and fine fractions ([less than or equal to]2000 [micro]m particles) having a chemical structure consistent with charcoal.

Acknowledgements

Funding provided by the Climate Change Research Program of the Australian Department of Agriculture, Fisheries and Forestry and by the Grains Research and Development Corporation are gratefully acknowledged. Soil sample collection activities conducted by all individual project groups within the SCARP is acknowledged and allowed the research to be applied to a diverse range of soils. Technical contributions from Kiralee Jury, Alana Massalsky, Christina Asanopoulos, and Andrea Ramirez Sepulveda are acknowledged.

References

Baldock JA, Skjemstad JO (2000) Role of the soil matrix and minerals in protecting natural organic materials against biological attack. Organic Geochemistry 31,697-710. doi: 10.1016/S0146-6380(00)00049-8

Baldock JA, Smemik RJ (2002) Chemical composition and bioavailability of thermally altered Pinus resinosa (Red pine) wood. Organic Geochemistry 33, 1093-1109. doi: 10.1016/S0146-6380(02)00062-1

Baldock JA, Oades JM, Nelson PN, Skene TM, Golchin A, Clarke P (1997) Assessing the extent of decomposition of natural organic materials using solid-state [sup.13]C NMR spectroscopy. Australian Journal of Soil Research 35, 1061-1084. doi: 10.1071/S97004

Baldock JA, Masiello CA, Grlinas Y, Hedges JI (2004) Cycling and composition of organic matter in terrestrial and marine ecosystems. Marine Chemistry 92, 39-64. doi: 10.1016/j.marchem.2004.06.016

Denef K, Six J, Bossuyt H, Frey SD, Elliott ET, Merckx R, Paustian K (2001) Influence of dry-wet cycles on the interrelationship between aggregate, particulate organic matter, and microbial community dynamics. Soil Biology & Biochemistry 33, 1599-1611. doi:10.1016/S0038-0717(01) 00076-1

Denef K, Six J, Merckx R, Paustian K (2004) Carbon sequestration in microaggregates of no-tillage soils with different clay mineralogy. Soil Science Society of America Journal 68, 1935-1944. doi:10.2136/ sssaj2004.1935

Forbes MS, Raison RJ, Skjemstad JO (2006) Formation, transformation and transport of black carbon (charcoal) in terrestrial and aquatic ecosystems. The Science of the Total Environment 370, 190-206. doi: l0.1016/j.scitotenv.2006.06.007

Golchin A, Baldock JA, Oades JM (1997) A model linking organic matter decomposition, chemistry and aggregate dynamics. In 'Soil processes and the carbon cycle'. (Eds R Lal, JM Kimble, RF Follett, BA Stewart) pp. 245-266. (CRC Press: Boca Raton, FL)

Grandy AS, Neff JC (2008) Molecular C dynamics downstream: The biochemical decomposition sequence and its impact on soil organic matter structure and function. The Science of the Total Environment 404, 292-307. doi: 10.1016/j.scitotenv.2007.11.013

Gregorich EG, Monreal CM, Schnitzer M, Schulten HR (1996) Transformation of plant residues into soil organic matter: Chemical characterization of plant tissue, isolated soil fractions, and whole soils. Soil Science 161, 680-693. doi:10.1097/00010694-19961000000005

Hamer U, Marschner B, Brodowski S, Amelung W (2004) Interactive priming of black carbon and glucose mineralisation. Organic Geochemistry 35, 823-830. doi: 10.1016/j.orggeochem.2004.03.003

Hammes K, Schmidt MWI, Smemik RJ, Currie LA, Ball WP, Nguyen TH, Louchouam P, Houel S, Gustafsson O, Elmquist M, Comelissen G, Skjemstad JO, Masiello CA, Song J, Peng P, Mitra S, Dunn JC, Hatcher PG, Hockaday WC, Smith DM, Hartkopf-Froeder C, Boehmer A, Luer B, Huebert B J, Amelung W, Brodowski S, Huang L, Zhang W, Gschwend PM, Flores-Cervantes DX, Largeau C, Rouzaud JN, Rumpel C, Guggenberger G, Kaiser K, Rodionov A, Gonzalez-Vila FJ, Gonzalez-Perez JA, de la Rosa JM, Manning DAC, Lopez-Capel E, Ding L (2007) Comparison of quantification methods to measure fire-derived (black/elemental) carbon in soils and sediments using reference materials from soil, water, sediment and the atmosphere. Global Biogeochemical Cycles 21, doi:10.1029/2006GB002914

Isbell RF (2002) 'The Australian Soil Classification.' Revised edn (CSIRO Publishing: Melbourne)

Jenkinson DS (1990) The turnover of organic carbon and nitrogen in soil. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 329, 361-368. doi:10.1098/rstb.1990.0177

Kogel-Knabner I (2002) The macromolecular organic composition of plant and microbial residues as inputs to soil organic matter. Soil Biology & Biochemistry 34, 139-162. doi:10.1016/S0038-0717(01) 00158-4

Marschner B, Brodowski S, Dreves A, Gleixner G, Gude A, Grootes PM, Hamer U, Heim A, Jandl G, Ji R, Kaiser K, Kalbitz K, Kramer C, Leinweber P, Rethemeyer J, Schaeffer A, Schmidt MWI, Schwark L, Wiesenberg GLB (2008) How relevant is recalcitrance for the stabilization of organic matter in soils? Journal of Plant Nutrition and Soil Science 171, 91-110. doi:10.1002/jpln.200700049

Parton WJ, Hartman M, Ojima D, Schimel D (1998) DAYCENT and its land surface submodel: description and testing. Global and Planetary Change 19, 35-48. doi:10.1016/S0921-8181(98)00040-X

Petersen BM, Berntsen J, Hansen S, Jensen LS (2005) CN-SIM--a model for the turnover of soil organic matter. I. Long-term carbon and radiocarbon development. Soil Biology & Biochemistry 37, 359-374. doi: 10.1016/j.soilbio.2004.08.006

Preston CM, Schmidt MWI (2006) Black (pyrogenic) carbon: a synthesis of current knowledge and uncertainties with special consideration of boreal regions. Biogeosciences 3, 397-420. doi: 10.5194/bg-3-397-2006

Schmidt MWI, Noack AG (2000) Black carbon in soils and sediments: Analysis, distribution, implications, and current challenges. Global Biogeochemical Cycles 14, 777-793. doi: 10.1029/1999GB001208

Skjemstad JO, Clarke P, Taylor JA, Oades JM, Newman RH (1994) The removal of magnetic materials from surface soils a solid state [sup.13]C CP/ MAS NMR study. Australian Journal of Soil Research 32, 1215-1229. doi: 10.1071/SR9941215

Skjemstad JO, Clarke P, Taylor JA, Oades JM, McClure SG (1996) The chemistry and nature of protected carbon in soil. Australian Journal of Soil Research 34, 251-271. doi: 10.1071/SR9960251

Skjemstad JO, Taylor JA, Smemik RJ (1999) Estimation of charcoal (char) in soils. Communications in Soil Science and Plant Analysis 30, 2283-2298. doi: 10.1080/00103629909370372

Skjemstad JO, Spouncer LR, Cowie B, Swift RS (2004) Calibration of the Rothamsted. organic carbon turnover model (RothC ver. 26.3), using measurable soil organic carbon pools. Australian Journal of Soil Research 42, 79-88. doi: 10.1071/SR03013

Smernik RJ, Oades JM (2000a) The use of spin counting for determining quantitation in solid state [sup.13]C NMR spectra of natural organic matter. 1. Model systems and the effects of paramagnetic impurities. Geoderma 96, 101-129. doi: 10.1016/S0016-7061(00)00006-9

Smernik RJ, Oades JM (2000b) The use of spin counting for determining quantitation in solid state [sup.13]C NMR spectra of natural organic matter. 2. HF-treated soil fractions. Geoderma 96, 159-171. doi:10.1016/S00167061 (00)00007-0

Smernik R, Baldock JA, Oades JM (2002) Impact of remote protonation on [sup.13]C CPMAS NMR quantitation of charred and uncharred wood. Solid State Nuclear Magnetic Resonance 22, 71-82. doi:10.1006/ snmr.2002.0065

Statsoft Inc (2007) 'STATISTICA (data analysis software system), version 8.0.' (StatSoft Inc.: Tulsa, OK) Available at: www.statsoft.com

Tan K (2003) 'Humic matter in soil and the environment.' (Marcel Dekker, Inc.: New York)

Wilson MA (Ed.) (1987) 'N.M.R. Techniques and applications in geochemistry and soil chemistry.' (Pergamon Press: Oxford, UK)

Zimmermann M, Leifeld J, Schmidt MWI, Smith P, Fuhrer J (2007) Measured soil organic matter fractions can be related to pools in the RothC model. European Journal of Soil Science 58, 658-667. doi: 10.1111/j.1365-2389.2006.00855.x

Zimmermann M, Bird MI, Wurster C, Saiz G, Goodrick I, Barta J, Capek P, Santruckova H, Smemik R (2012) Rapid degradation of pyrogenic carbon. Global Change Biology 18, 3306-3316. doi:10.1111/j.13652486.2012.02796.x

J. A. Baldock (A,B), J. Sanderman (A), L. M. Macdonald (A), A. Puccini (A), B. Hawke (A), S. Szarvas (A), and J. McGowan (A)

(A) CSIRO Land and Water/Sustainable Agriculture Flagship, PMB 2, Glen Osmond, SA 5064, Australia.

(B) Corresponding author. Email: jeff.baldock@csiro.au

Table 1. Properties of the five soils included in the comparison of
manual and automated soil carbon fractionation  methodologies

Sample         Location           Depth    Classification
                                   (cm)

SS6            Tallagalla, Qld     0-5     Dermosol
SS7            Toowoomba, Qld      0-5     Ferrosol
SS8            Waco, Qld           0-10    Vertosol
SS10           Gympie, Qld         0-10    Chromosol
Weisenboden    Adelaide, SA        0-10    Vertosol

Sample         [pH.sub.w]     OC content         Clay
                  (A)       (g [kg.sup.-1])   content (%)

SS6               6.6            71.4             45
SS7               5.9            143.0            16
SS8               8.2            28.0             61
SS10              5.0            44.4             21
Weisenboden       7.7            31.0             40

(A) Determined on a 1 : 5 soil: water suspension.

Table 2. Chemical shift regions into which the [sup.13]C NMR signal
intensity was divided, the proposed major types of organic carbon (OC)
associated with each region, and the calculations used to allocate
total signal intensity to each type of OC

Proportional allocations for each region were calculated by dividing
the amount of signal intensity obtained from each region by the total
signal intensity found in the 300 to -100 ppm chemical shift region

Chemical       Proposed dominant       Calculations used to quantify
shift          type of OC              the amount of signal intensity
limits                                 associated with each type of
(ppm)                                  carbon

290-265        Carbonyl and amide
                 spinning side band
265-245        O-aryl spinning side
                 band
245-215        Aryl spinning side
                 band
215-190        Ketone C                (215-190 ppm)
190-165        Carbonyl and amide C    (190-165 ppm)+2 x (290-265 ppm)
165-145        O-aryl C (phenolic      (165-145 ppm)+2 x (265-245 ppm)
                 and furan)
145-110        Aryl and                (145-110 ppm)+2 x (245-215 ppm)
                 unsaturated C
110-90         Di-O-alkyl C            (110-90 ppm)
90-05          O-alkyl C               (90-05 ppm) - (290-265 ppm)
65-45          Methoxyl and            (65-45 ppm) - (265-245 ppm)
                 N-alkyl C
45-0           Alkyl C                 (0-45 ppm) - (245-215 ppm)
300 to -100    Total signal
                 intensity

Table 3. Proportions of lignin and charcoal C allocated to the aryl
and O-aryl [sup.13]C NMR spectral regions

                  Aryl           O-aryl
                  (110-145ppm)   (145-160ppm)

Lignin            0.32           0.18

ROC (poly-aryl    0.72           0.15
char like OC)
                  Source

Lignin            Average allocations obtained for spruce and red
                    alder lignin (Wilson 1987) and Sigma Aldrich
                    alkali lignin (Product no. 370959)
ROC (poly-aryl    Average of 10 [sup.13]C NMR spectra obtained for a
char like OC)       variety of different types of charcoal
                    (Baldock et al. 2004;
                    J. O. Skjemstad, unpubl. data)
COPYRIGHT 2013 CSIRO Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2013 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Baldock, J.A.; Sanderman, J.; Macdonald, L.M.; Puccini, A.; Hawke, B.; Szarvas, S.; McGowan, J.
Publication:Soil Research
Article Type:Report
Geographic Code:9TAIW
Date:Oct 1, 2013
Words:10133
Previous Article:Foreword.
Next Article:Predicting contents of carbon and its component fractions in Australian soils from diffuse reflectance mid-infrared spectra.
Topics:

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters