Department for Environment, Food and Rural Affairs
Research project final report
Project title / Exploration of methodologies for accurate routine determination of soil carbon
Sub-Project iv of Defra Project SP1106: Soil carbon: studies to explore greenhouse gas emissions and mitigation
Defra project code / SP1106
Contractor organisations / SKM Enviros
Cranfield University
Centre for Ecology and Hydrology
British Geological Survey
Rothamsted Research / North Wyke
Report authors / Phil Wallace (), Guy Kirk, Pat Bellamy, Bridget Emmett, David Robinson, Inmaculada Robinson, Barry Rawlins, Ron Corstanje, Roland Bol.
Project start date / October 2010
Sub-project end date / March 2011

Exploration of methodologies for accurate routine determination of soil carbon

Sub-project iv of Defra project SP1106: Soil carbon: studies to explore greenhouse gas emissions and mitigation

Contents

Executive summary

1 Introduction

2 Review of sampling and analysis for soil carbon

2.1 Identification and exploration of the different soil carbon fractions

2.2 Review of sampling strategies

2.3 Review of laboratory analytical techniques (UK and international)

3 Workshop discussion outcomes and SOPs (see annexes)

4 Review of New and emerging technologies

Annex 1 Soil sampling Standard Operating Procedure

Annex 2 Soil carbon determination Standard Operating Procedure

Executive Summary

There has been a number of soil monitoring exercises carried out over the last 35 years in England and Wales, as well as other areas of the UK and internationally. The methods used at each survey for soil sampling and soil carbon determination in England and Wales have not been consistent, which has led to difficulties in identifying and quantifying trends in changes in soil carbon over time. Soils contain a large stock of carbon and any changes, due to e.g. changes in land management, may be significant when compared to the UK’s annual emissions and therefore need to be accurately determined. This project was therefore tasked with standardising the methodology for future monitoring schemes through the production of standard operating procedures agreed by the UK’s soil science community. The project also explored the various soil carbon fractions, their definitions and relationships with models. In the future, soil carbon determination may be by the use of new technologies and these were reviewed.

Soil organic matter in soil may survive intact for a relatively short time period of days to a few years, an intermediate time period of years to decades, or be recalcitrant and remain in the soil for decades to centuries. Methods to measure these pools of soil carbon have been developed using physical, chemical or biological fractionation, or combinations of these methods.

Soil sampling methodologies were extensively reviewed by Black et al. (2008). Sampling should be carried out such that determinations of both soil carbon stock and change can be carried out. This requires an adequate number of sample replicates to be taken, bulked into composite samples and sub-sampled, if appropriate, from an area of land identified as suitable. Soils have to be sampled to a depth that is consistent with past measurements whilst allowing for changes in techniques in the future. Soil bulk density in the topsoil, but also by horizon for soil carbon stocks, needs to be measured. Issues were discussed and agreed at a workshop by the soil science community and a standard operating procedure proposed.

Soil carbon analysis laboratory methods were reviewed. Traditionally, wet and dry combustion methods have been used such as a modified Walkley Black or loss on ignition. Total organic carbon by combustion, measuring CO2 evolution at high temperature with an elemental analyzer after acid pre-treatment of the soil, has also been more recently utilised. These methods, as well as future technologies, were discussed at the workshop and agreement reached as to the methods to be used in future soil monitoring schemes based on the more traditional methods. A standard operating procedure was agreed.

Nine new and emerging technologies were reviewed for their potential to monitor soil carbon. Visible/near and mid infra-red diffuse reflectance spectroscopy show promise as well as thermo-gravimetry – differential scanning calorimetry and Rock-Eval pyrolysis.

1 Introduction

Monitoring of soils in England and Wales has been carried out in two samplings of the National Soil Inventory (originally in the period 1978-1983 followed by re-sampling in phases between 1994-1995, 1995-1996 and 2003) and the samplings of the Countryside Surveys in 1990, 2000 and 2007. Other surveys in the UK have been carried out in Scotland, Northern Ireland, of forested areas and for geochemical properties, including soils, since the 1960s. Soil monitoring has been carried out in other parts of the EU and elsewhere in the world.

The sampling and analytical methods employed during the soil monitoring of England and Wales have not been standardised leading to potential differences such that changes in organic matter in soils over time are difficult to detect. In order to measure changes in soil carbon in the future in response to impacts such as climate change and management practices, standardisation of techniques is required. Soils in the UK contain about 10 billion tonnes of carbon, equivalent to over 50 times the UK’s annual greenhouse gas emissions, so small changes in soil carbon due to changes in soil management practices or land use may have a large impact on our emissions. Soil monitoring in the future should be designed to be able to detect changes in soil carbon from past surveys such that soil carbon levels can be adequately monitored.

Soil organic matter includes fresh plant and animal residues, decomposed materials in more resistant forms and elemental carbon. These fractions require definition, especially in relation to the pools in soil carbon models. Traditional methods of soil carbon measurement include the destruction of the organic matter by chemical or combustion techniques. Other, non-destructive, techniques for soil organic matter determination have been investigated, including in situ and ex situ methods.

The aims of this sub-project were:

1 To review methodologies, in UK and internationally, for determining soil carbon and its various fractions, including field sampling and laboratory analysis.

2 To define a new standard operating procedure (SOP) for use in future soil surveys, discussed and agreed through a workshop of experts from the soil science community.

3 To explore new and emerging techniques in terms of their robustness across different soil types, practicality for non-specialists, the time to develop and adopt them, and their cost-effectiveness.

2 Review of sampling and analysis for soil carbon

2.1 Identification and exploration the different soil carbon fractions

2.1.1 Measures of SOM quality for monitoring and modelling

The decomposability or ‘quality’ of soil organic matter (SOM) as a biological substrate varies from labile material that decomposes within a few weeks or months, to recalcitrant material that can persist for centuries. Factors governing SOM quality include its chemical composition, its stabilisation by association with clay minerals and oxides and other mechanisms, and its location within the soil as this influences access of microbes, oxygen and other reactants to it (Sollins et al., 1996; Six et al., 2002; von Lützow et al., 2006). As a result, SOM quality varies more-or-less continuously across a spectrum, and discrete, operationally-defined ‘fractions’ can only describe this continuous variation approximately. Nonetheless, some form of operationally-defined fractionation is necessary for making measurements to characterize SOM for monitoring and modelling. The plethora of possibilities available for this is illustrated in Table 1.

Table 1 Soil organic matter pools defined by turnover times, and related measured fractions (after Wander, 2004)

Pools / Measured fractions
Labile SOM
Half-life days to a few years
Material of recent origin or living components of SOM; material of high nutrient or energy value; physical state or location does not impede access to microbes, oxygen or other reactants / Microbial biomass
Chloroform-labile SOM
Microwave-irradiation-labile SOM
Amino compounds
Phospholipids
Labile substrates
Mineralizable C or N, estimated by incubation
Substrate-induced activity
Soluble, extractable by hot water or dilute salts
Easily oxidized by permanganate or other oxidants
Residues for which chemical formula can be
described, inherited from living organisms
Litter, vegetative fragments or residues
Non-aggregate-protected SOM
Polysaccharides, carbohydrates
Intermediate SOM
Half-life of a few years to decades
Physical state or location impedes access / Partially-decomposed residues and decay products
Amino compounds, glycolproteins
Aggregate-protected SOM
Others
Acid/base hydrolyzable humic substances
Mobile humic acids
Recalcitrant SOM
Half-life of decades to centuries
Recalcitrant because of chemical composition and/or mineral association / Refractory compounds of known origin
Aliphatic macromolecules (lipids, cutans, algaenans,
suberans)
Charcoal
Sporopollenins
Lignins
Others
High molecular weight, condensed SOM
Humin
Non-hydrolyzable SOM
Fine-silt, coarse-clay associated SOM

Most existing SOM models are based on a number of pools (usually two for plant litter and three for SOM) defined by first order decomposition rate constants, and SOM turnover is calculated from transformations between the pools (Jenkinson, 1990; Parton et al., 1993; Bruun et al., 2010). Such models have been widely and successfully used to simulate changes in total soil carbon in response to changes in soil management and land use. However, they have the practical limitation that, because the pools are hypothetical, and not directly measurable, their parameters must be fitted to data. But the calibration necessarily cannot cover all soils and land uses; notably, most current models do not cover the cold, wet, humose soils where the largest carbon stocks occur (Bellamy et al., 2005; Schultze & Freibauer, 2005). The models also have the limitation that the fitted, discrete pools are a crude representation of continuously varying SOM quality, and they are therefore poor at revealing underlying processes and mechanisms (Bruun et al., 2010). Also, the number of parameters to be fitted for a multiple pool model – including those for the dependences of the rate constants on temperature, moisture, redox and other variables – is very large and it may be difficult to fit them with much certainty.

There has been some success in developing measurement schemes that approximate to particular model pools (Hassink, 1995; Trumbore & Zheng, 1996; Magid et al., 1996; Six et al., 2000; Christensen, 2001; Sohi et al., 2001; Zimmermann et al., 2007). But the fundamental problem remains that the pools and corresponding fractions are hypothetical. This has led to calls to base models on explicitly measurable pools with variable reactivities (Hassink, 1995; Christensen, 1996; Elliott et al., 1996; Arah & Gaunt, 2001) and on continuously varying measures of SOM quality, such as particle size, particle density or resistance to oxidation (Ågren & Bosatta, 1996; Bruun et al., 2010).

2.1.2 Measurement schemes

Physical fractionation

Separation by physical fractionation procedures emphasises the role of physical protection and location in SOM turnover (Balesdent, 1996; von Lützow et al., 2007). Also these procedures are often easy and cheap in comparison with chemical or biological methods.

Particle size

Particle size reflects soil mineralogy and chemical composition, and these are important determinants of interactions between SOM and mineral matter; for example, sand-sized quartz grains are relatively inert compared with clay-sized minerals. Hence fractionation based on particle size is a reasonable basis for separating SOM of different quality. Figure 1 (from von Lützow et al., 2007) summarises typical distributions of SOM across particle size fractions in temperate soils, and the corresponding distributions of measures of SOM quality and turnover rates. It shows a near-continuously increasing proportion of SOM associated with decreasing particle size. However, the turnover times of SOM in different particle-size fractions, gauged by stable isotope fractionation, are very variable (von Lützow et al., 2007). Particle size and related interactions with SOM will vary continuously from coarse to fine particles. Better relationships between particle size and turnover time may be obtained by separating a size continuum to resolve the distribution better, possibly in combination with other quality variables (Bruun et al., 2010). Traditional methods used to separate SOM according to particle size include sieving and sedimentation, sometimes after dispersion with ultrasonic vibration or a chemical dispersant (Christensen, 1992). Bruun et al. (2010) review methods from colloid science that might be adopted to measure continuous distributions of SOM in silt and clay fractions of soils, and conclude several of the methods are promising.

So called ‘particulate organic matter’ (POM) is used as an indicator of early trends in SOM and nutrient status in managed soils (Magid et al., 1996; Gregorich & Carter, 1997; Yakovchenko et al., 1998; Carter, 2002). It is separated from coarse (> approx. 50 μm) particle size fractions by sieving and sedimentation (Cambardella & Elliott, 1993; Salas et al., 2003).

Figure 1 Typical distributions of SOM across particle size fractions in temperate soils, and the corresponding distributions of measures of SOM quality and turnover rates (von Lützow et al., 2007)

Density

Fractionation methods based on density generally produce two fractions – heavy and light – generally by centrifugation in a heavy liquid (Christensen, 1992). The scheme of Sohi et al. (2001) uses a sequence of separations in a concentrated sodium iodide solution, with and without ultrasonic dispersion, to obtain four fractions: a dissolved fraction; a light fraction, taken to be free organic matter; a second light fraction, taken to be intra-aggregate organic matter; and a heavy fraction, which is the residual organo-mineral matter. The fractions display consistent differences in their chemical properties, as determined by NMR spectroscopy, pyrolysis mass spectrometry and thermal analysis (Lopez-Capel et al., 2005b; Poirier et al., 2005; Sohi et al., 2001, 2005); these properties may correlate with in situ SOM quality. However, further work is needed to calibrate this indirect measure of SOM quality with the soil conditions at the time of sampling. Sequential density fractionation has been made with up to eight different densities (Baisden et al., 2002; Sollins et al., 2009), but this is a tedious procedure. An alternative may be to use density gradient centrifugation, in which a density distribution of particles is created and samples withdrawn at different depths within it (Bruun et al., 2010).