Soil organic carbon content is normally measured using some form of chemical oxidation (e.g. Walkley and Black 1934) or by dry combustion (e.g. Leco carbon determinator, Merry and Spouncer 1988) as total carbon. TOC values, however, may not adequately describe the role of carbon in many soil processes, and it may be more appropriate to partition it into several pools or fractions with varying degrees of reactivity or biological stability (van Veen et al. 1984; Parton et al. 1987; Jenkinson and Coleman 1994). These pools can be modelled with computer simulations, and range flora labile organic matter derived from plant debris, to moderately to highly resistant carbon from humified organic matter, and inert or highly protected organic matter (Skjemstad et al. 1998).
For carbon modelling purposes, it would be useful to have a set of defined soil fractions that could be closely related to the conceptual pools used in particular carbon turnover models. Skjemstad et al. (1996) showed that a significant proportion of soil organic carbon in most Australian soils was present as charcoal (char) and that this fraction probably represented a soil pool which was highly resistant to microbial decomposition. This fraction can be determined routinely using high-energy ultraviolet (UV) photo-oxidation and [sup.13]C NMR spectroscopy and was considered to be a good candidate as a useful soil pool for modelling purposes. This fraction also appeared to fit well with the RothC model for C turnover (Jenkinson and Coleman 1994), which contained an inert pool as part of its pool structure. Subsequently, Skjemstad et al. (2004) used the char-C fraction along with the >53[mu]m particulate organic carbon (POC) fraction reported by Cambardella and Elliot (1992) to initialise, calibrate, and verify the RothC model. In this model, the POC fraction is equated to the resistant plant material pool, char-C to the inert pool, and the difference between total organic carbon (TOC) and char-C plus POC pools to the humic pool.
The TOC and POC fractions can be measured easily, but the char-C analytical procedure cannot. The soil must be separated to isolate the <53 [mu]m size fraction which normally contains >90% of the char, which is then subjected to high-energy UV photo-oxidation to remove labile and humified OC, and treated with HE The residual fraction is then analysed by [sup.13]C nuclear magnetic resonance (NMR) spectroscopy to determine its char-C content (Skjemstad et al. 1999). NMR analysis is required because the material remaining after photo-oxidation represents a mixture of char and other organic carbon which is strongly protected by inclusion within microaggregates against the photo-oxidation process. The latter is usually very small, but with some soils can represent up to 50% of the remaining OC and would therefore cause an overestimation of the char-C content if NMR analysis was not performed. This fractionation scheme proposed by Skjemstad et al. (1999) is therefore very time-consuming and expensive due to the difficulty in separating enough material to obtain an acceptable NMR spectrum (~50 mg C).
For the routine analysis for carbon fractions of a large number of soils, a simpler and cheaper method of acceptable accuracy is needed. The fractionation scheme discussed above relies on the availability of photo-oxidation equipment, access to a solid-state [sup.13]C NMR spectrometer, and hazardous chemical treatments with HF, and is also laborious. Infrared spectroseopy offers a simple, rapid, and low-cost alternative, with a further advantage that it is sensitive to the chemistry of both organic and mineral components in the soil. MIR spectroscopy is characteristised by fundamental vibrations of molecules associated with particular chemical functional groups. For example, the technique enables the identification of specific soil minerals and of organic matter functional groups such as alkyl, carboxylic (protonated and non-protonated), carbohydrates, amide, amine, and most importantly aromatic functional groups (Van der Marel and Beutelspacher 1976; Skjemstad and Dalal 1987; Theng and Tate 1989; Theng et al. 1992; Piccolo 1994; Janik and Skjemstad 1995; Wander and Traina 1996; Janik et al. 1998).
The advantages of MIR spectroscopy, particularly when combined with multivariate chemometric techniques such as partial least-squares (PLS) regression for the prediction of sample properties, are its analytical speed and simplicity. No fractionation is required, hazardous chemical reagents are avoided, and many soil components can be predicted from a single spectrum, although air-drying and grinding is advantageous. The MIR-PLS technique has already been shown to be suitable for the prediction of TOC from spectra of whole soils (Janik and Skjemstad 1995; Janik et al. 1998) and because it is sensitive to the various organic functional groups contributing to soil organic matter should, in principle, be sensitive to the distribution of different carbon pools, offering a realistic and practical approach for large-scale carbon pool analysis. The increased potential of MIR spectroscopy has resulted from the widespread availability of PLS software, which provides a powerful and robust quantification tool (Holmgen and Norden 1988; Janik and Skjemstad 1995; Janik et al. 1998; Hazel et al. 1997).
The procedure for PLS analysis adopted here closely follows that detailed by Haaland and Thomas (1988), and later by Janik and Skjemstad (1995) and Janik etal. …
No comments:
Post a Comment