Managing the soil-plant system to mitigate

Managing the soil-plant system to mitigate

Managing the soil-plant system to mitigate atmospheric CO2 Discussion paper for the Soil Carbon Sequestration Summit January 31-February 2 2011 Compiled by Uta Stockmann Contributors: Mark Adams John Crawford Damien Field Nilusha Henakaarchchi Meaghan Jenkins Alex McBratney Vivien de Remy de Courcelles Kanika Singh Uta Stockmann Ichsani Wheeler © The University of Sydney, Faculty of Agriculture, Food and Natural Resources The United States Studies Centre at the University of Sydney Soil Carbon Initiative | Dow Sustainability Program | Alcoa 2

Managing the soil-plant system to mitigate

Table of Contents 1 Introduction ___ 4
2 Soil – A terrestrial pool for organic carbon ___ 5
2.1 Effects of climatic conditions and ecosystem conditions on SOC pools ___ 6
Climatic conditions ___ 6
Ecosystem conditions (land use change ___ 6
2.2 Environmental conditions affecting the decomposition of SOC ___ 7
2.3 Biological processes affecting the decomposition of SOC ___ 7
Priming effects ___ 7
Biodiversity ___ 8
Roots and root exudates ___ 8
2.4 Chemical and physical processes affecting the decomposition of SOC ___ 9
3 Soil carbon measurements – Pools and fluxes ___ 12
3.1 SOC pools ___ 12
Measurement of SOC ___ 12
SOC fractionation ___ 13
Imaging SOC in situ with structure ___ 14
3.2 C fluxes ___ 17
Measurement of C fluxes ___ 17
4 Soil carbon modelling ___ 19
4.1 Process oriented versus organism-oriented models ___ 19
Model characteristics ___ 19
Model limitations ___ 25
4.2 Other models used to predict SOM stocks, content, storage or change ___ 28
Empirical regression models ___ 28
Landscape models ___ 28
The whole systems modelling approach ___ 29
4.3 Where to from here in modelling of SOM dynamics ___ 30
Prospects for a potential Next Generation Soil Carbon Model ___ 30
5 Soil carbon management – Enhancing soil carbon ___ 34
5.1 Purpose of soil carbon management practices ___ 34
5.2 Agricultural practices to enhance SOC ___ 35
Conservation farming ___ 35
Biological farming ___ 36
Biochar ___ 39
Managed forest systems ___ 41
5.3 Soil C management - SCS as a valid CO2-equivalent credit ___ 41
5.4 Short term and long term trends in soil C ___ 44
6 Conclusions ___ 45
Soil Carbon Initiative | Dow Sustainability Program | Alcoa 3

Managing the soil-plant system to mitigate

Soil Carbon Initiative | Dow Sustainability Program | Alcoa 4 1 Introduction Approximately 9 Gt of C are emitted to the atmosphere each year on a global scale by anthropogenic sources. However, the atmospheric increase has only been in the order of 3.5 Gt C yr-1 , which highlights the important regulatory capacity of global C pools. In this context, soil organic carbon and its potential to become a ‘managed’ sink for atmospheric CO2 has been discussed widely in the scientific literature (i.e. Chabbi and Rumpel, 2009, Guo and Gifford, 2002, Kirschbaum, 2000, Lal, 2004a, Lal, 2004b, Lal, 2008, Luo et al., 2010, Post and Kwon, 2000, Smith, 2008).

However, to reward this ‘good’ management of the soil carbon pool, the following fundamental questions should be considered when addressing the potential of the soil-plant system to ‘sequester’ organic carbon1 : What is the overall purpose of soil carbon sequestration? For instance, how do we define soil carbon sequestration (SCS) and how do we consider requirements of permanence?

Does all carbon in the soil have the potential to behave in the same way, or are there different pools associated with distinct behaviour? Is there a pool of carbon that is unavailable for the microbial community and so does not contribute to function? What is the value in increasing easily decomposable soil organic matter (SOM) aside from its sequestration benefits? For how long must this increase be maintained to be considered as SCS? Consequently, what is more important – long-term SCS or the functioning of the soil? And therefore, are both equally possible?

How do we accomplish/support this paradigm as a science community? How should we promote the benefits of SCS to policy makers/farmers/landholders (i.e.

the potential of SCS to mitigate climate change, the use of SCS as a platform for sustainable agriculture) and how do we follow through with suitable answers to questions such as measurement, modelling, monitoring and permanence for SCS and/or management advice supported by research? Do we need to improve models of SOM dynamics in order (i) to understand the functioning of the soil ecosystem better and in order (ii) to better assist farmers with management decisions? Can both these questions be addressed with the same model? This review will summarize the state of the art in soil carbon research and will highlight potential or recommended future research approaches.

It is primarily intended however, to stimulate discussion and discourse in this important topic area with a view to addressing these pressing questions with research that is directly linked to real world outcomes. 1 Soil carbon sequestration (SCS): implies an increase in the managed soil carbon pool for a defined period where the increased carbon is sourced from atmospheric CO2.

2 Soil – A terrestrial pool for organic carbon In a global context, the quantity of carbon stored in soil is second only to that in the ocean (38400 Gt) and represents the largest store of terrestrial organic carbon, at more than four times the biotic carbon pool (which has an estimated amount of 560 Gt of organic carbon) (Figure 1). Approximately 2344 Gt of organic carbon is stored in the top three meters of soil, with 54% or 1500 Gt of organic carbon stored in the first meter of soil and about 615 Gt stored in the top 20 cm (Guo and Gifford, 2002, Jobbagy and Jackson, 2000). We know broadly that plant production and patterns of allocation determine the relative distribution of carbon with soil depth (Jobbagy and Jackson, 2000).

It is believed that older organic matter materials are stored deeper in the soil profile. For instance, a review by Trumbore (2009) postulated increased ages of low-density C and microbial phospholipid acids with depth. Whilst Fontaine et al. (2007) proposed an increase of mean residence times of SOC of up to 2000-10000 years for a depth beyond 20 cm which is most likely an indication for reduced soil organic carbon (SOC) turnover at depth resulting from changes in microbial activity and substrate supply. They also detected an increase of organo-mineral complexes with depth, however, no change of the chemical composition of SOM was detected in the soil profiles studied.

Figure 1 Total annual flux of carbon in gigatons (billions of tonnes) through the most biologically active pools (as compared to the deep-ocean and lithosphere). Natural annual input of ‘new’ C to the biosphere through volcanism and rock dissolution is balanced by long term burial of C in ocean sediments. Annual anthropogenic input of ‘new’ C into the atmospheric pool (averaged over the last 40 years) comes from land clearing (1-2 Gt annually) and fossil fuel sources (8 Gt annually). Estimates adapted from Volk (2008). Soil Carbon Initiative | Dow Sustainability Program | Alcoa 5

2.1 Effects of climatic conditions and ecosystem conditions on SOC pools Climatic conditions Worldwide, soil C storage generally increases as mean annual temperature decreases (Post et al., 1982) and cool/cold climate regions are characterized by their carbon-rich soils.

For example, the arctic and boreal ecosystems of the northern hemisphere contain a large proportion of the world’s soil C, with estimates ranging from 90-260 Gt of C for upland boreal forest soils, 120-460 Gt of C for peatland soils and 60-190 Gt of C for arctic tundra soils (Hobbie et al., 2000).

A change in just 10 % of soil organic carbon pool would be equivalent to 30 years of anthropogenic emissions and could dramatically affect concentrations of atmospheric CO2 (Kirschbaum, 2000). Any enhancement of concentrations of greenhouse gases in the atmosphere is set to accelerate the rate of warming that may in turn change net primary productivity (NPP), inputs of C to soil, soil microbial activity and eventually the release of further CO2 from soil. Kirschbaum (2000) concluded that global warming is likely to reduce soil organic carbon by stimulating decomposition rates whilst simultaneously increasing soil organic carbon through enhanced net primary production resulting from increased CO2 levels.

However, the net effects of changes in soil organic carbon on enhancing atmospheric CO2 levels were predicted to be small over the next centuries.

Ecosystem conditions (land use change) A review by Guo and Gifford (2002) highlights the influence of land use changes on soil C stocks. Meta-analysis of 74 publications indicated that land use changes from pasture to plantation (-10 %), native forest to plantation (-13 %), native forest to crop (-42 %) and pasture to crop (-59 %) decreased total carbon stocks whereas land use changes from native forest to pasture (+8 %), crop to pasture (+19%), crop to plantation (+18 %) and crop to secondary forest (+53 %) increased total C stocks. We can conclude that land use change from cropland to pasture or cropland to permanent reforestation are generally the ‘most’ efficient aggrading systems in terms of soil organic carbon (SOC) gain whereas the conversion to cropland or monocultures are common examples of degrading systems.

There are major and well-known differences in the quantity, quality and timing of organic matter inputs to soil that result from variation in species composition within community types (i.e. relative abundance of N-fixing species) or from wholesale changes in community phenotype (i.e. cropland, grassland, shrubland, woodland, forest) that may result from either management or variation in edaphic conditions at a local scale (i.e. Binkley and Menyailo, 2005, Hart et al., 2005). Recent research investigated the long-term (1930-2010) dynamics of SOC after land use change in Java, Indonesia (Minasny et al., 2010).

It is well known that soils in tropical environments generally lose SOC due to conversion of primary forest to plantation and cultivated lands. However, in Java, total SOC stocks increased after 1970 most likely resulting from human influences and management practices (i.e. the ‘GreenSoil Carbon Initiative | Dow Sustainability Program | Alcoa 6

Revolution’) with a net accumulation rate in the first 10 cm of 0.2-0.3 Mg C ha-1 yr-1 during the period of 1990-2000. 2.2 Environmental conditions affecting the decomposition of SOC Litter decomposition is the major pathway of carbon release from soils. The most important controls over the rate of decomposition are climate, moisture and temperature, litter quality, and soil microbial community composition (Meentemeyer, 1978, Melillo et al., 1982, Parton et al., 2007). Litter inputs provide the carbon source for microbial growth and activity, the higher the quality (i.e. more readily decomposable) the more rapidly the microbial community can grow (Agren and Bosatta, 1996).

Decomposition has been described as a key or even ‘bottleneck’ process in carbon and nutrient cycling. The chemical composition of decomposing material, especially characteristics such as C:N ratios and lignin content, are crucial to determining how quickly decomposition proceeds (Melillo et al., 1982, Meentemeyer, 1978). In general, litter decomposition is negatively related to C:N ratios, lignin content and lignin:N ratios, and positively related to N concentrations (Melillo et al., 1989, Melillo et al., 1982). 2.3 Biological processes affecting the decomposition of SOC Priming effects Priming effects on soils were first recorded more than 50 years ago (Bingeman et al., 1953) and a recent review defined soil priming as “strong short-term changes in the turnover of soil organic matter caused by comparatively moderate treatments of the soil” (Kuzyakov et al., 2000).

The existence of ‘priming effects’ of added C (or N) on rates of mineralisation of soil organic matter is well documented (i.e. Kuzyakov et al., 2000, Fontaine et al., 2003, Sayer et al., 2007). Priming effects can be negative or positive. In particular, addition of purified labile C substrates such as glucose (Dilly and Zyakun, 2008), fructose and oxalic acid (Hamer and Marschner, 2005), or cellulose (Fontaine et al., 2007) have been shown to clearly stimulate mineralisation of organic matter. One theory of the priming effect is that of cometabolism – additions of labile or fresh carbon stimulate growth of a suite of microorganisms that in turn leads to an increase in microbial enzyme production (Kuzyakov et al., 2000, Hamer and Marschner, 2005).

Fontaine et al. (2003) go onto suggest that changes in land use and agricultural practices that increase the distribution of fresh carbon at depth could stimulate the mineralisation of ancient buried carbon. For instance, would a land use change from shallow to deep-rooted grasses therefore have an overall negative feedback on SOC levels due to the potential release of C buried at depth? Or would the increased allocation of C below ground outweigh this potential effect? Change in biomass allocation may therefore have a potential impact on the whole soil C storage system. However, it is still to be proven if all soil microorganisms follow the same laws or if possible inherent characteristics affect microbial diversity and SOC sequestration (Chabbi and Rumpel, 2009).

Soil Carbon Initiative | Dow Sustainability Program | Alcoa 7

Biodiversity It is still discussed in the scientific literature to what extent diversity and certain plant species affect soil biota and SOM dynamics. Generally, it is believed that increased plant diversity increases SOC levels. It is also postulated that biodiversity promotes ecosystem stability (i.e. feedback mechanisms such as better adjustment to climatic conditions or pest outbreaks). However, mathematical studies on the stability of species communities led to the assumption that communal complexity may result in instability (chaos). Recent research suggested that the ‘organization’ of diversity most likely determines stability (i.e.

simple or complex species interactions in the agroecosystem) (i.e. Susilo et al., 2004). There have been numerous studies of litter decomposition, mostly focusing on the influence of single plant species and mixtures of leaf litter on decomposition (Melillo et al., 1982, Hobbie, 1996, Chapman et al., 1988, O'Connell, 1986, Briones and Ineson, 1996, McTiernan et al., 1997, Xiang and Bauhus, 2007). Individual plant species are known to influence soil microbial activity and nutrient cycling through the quality and quantity of organic matter return to the soil (Binkley and Menyailo, 2005, Galicia and García-Oliva, 2004).

The presence/absence of plant species that host N-fixing bacteria (such as legumes) or root systems with mycorrhizal associations, often enhance nutrient uptake and can provide a pathway for return of carbon substrate to microbes/soil (Hobbie, 1992). Roots and root exudates Root exudates are an important C input for the soil, representing 5-33 % of daily photoassimilate (Chabbi and Rumpel, 2009). Root–derived organic matter inputs are chemically diverse ranging in complexity from readily decomposable substrates, such as soluble sugars, amino acids and organic acids, to substrates from root turnover that require greater energy investment to decompose.

Exudates fall into two categories: exudates which are a result of passive diffusion and over which the plant has little control and secondly exudates which are released for a specific purpose and over which the plant exerts a certain degree of control (Paterson et al., 2007, See reviews by Jones et al., 2004). It has been widely assumed that easily degradable plant exudates are almost exclusively degraded by bacteria (see reviews by Jones, 1998), whilst fungi play an important role in the degradation of recalcitrant organic materials, such as lignin and also dominate the decomposition of cellulose and hemi-cellulose (reviewed by Boer et al., 2005).

There have been assumptions that roots exudates contribute to the depletion of SOC stocks through the ‘rhizosphere priming effect’ and it has also been postulated that overall rates of SOC decomposition increased up to 5 fold in response to root exudates (see Sanderman et al., 2010 for examples).

Soil Carbon Initiative | Dow Sustainability Program | Alcoa 8

2.4 Chemical and physical processes affecting the decomposition of SOC Research on SOM ultimately led to the assumption that there is a refractory (stable) component within SOM. The ongoing discussions regarding the chemical and physical processes determining the composition of SOM have been very well summarized in a recent review by Kleber and Johnson (2010) and are highlighted in the following. (a) The view of recalcitrant substances being a part of SOM has been shaped by the solubility of SOM when subjected to chemical (alkaline/acid) extractions Chemical extraction procedures leave residues behind which are assumed to be resistant (recalcitrant) because of their complex polymeric macromolecular structure.

This observation led to the development of the humus concept which postulates that during decomposition processes, SOM is humified into humic substances with different turnover times, i.e. fulvic acids (decades/centuries), humins and humic acids (millennia) (Schlesinger, 1977). However, this concept of inherent chemical stability is questioned in the literature based on the theory that the extracted humic substances are most likely a product of the extraction procedure rather than a true in-situ component of SOM. For instance, the latest technologies such as NMR and synchrotron spectroscopy did not detect evidence for discrete humic molecules in unprepared soil (Lehmann et al., 2008b).

Gleixner et al. (1999) showed that instead of macromolecules, substances with lower molecular weight (proteins or peptides) are more likely to be preserved in soil during decomposition and humification processes. They studied the chemical structures involved in soil organic carbon sequestration applying Curie point pyrolysis-gas chromatography coupled on-line to mass spectrometry (Py-GC/MS) and isotope ratio mass spectrometry (Py-GC IRMS). Moreover, the assumed existence of a high proportion of aromatic C in humic materials has also been questioned (Gleixner et al., 2002). These new findings provide support for the view that microbial activity may be the primary active agent for SOM stabilization, and that most likely it is the constituent C after integration into new microbial derived molecules that remains in the soil and not the precursor substance itself (see Chabbi and Rumpel, 2009).

(b) The view on decomposition rates and humification pathways lead to the distinction of SOM in terms of soil function (as conceptualised in process-oriented models of SOM dynamics) As shown in Figure 2 non-living SOM is divided into at least three C pools (1) the active pool with turnover rates of years (root exudates, rapidly decomposed components of fresh plant litter) (2) the intermediate or slow pool with turnover rates of centuries and (3) the passive pool with turnover rates of millennia (stabilized organic matter due to chemical or physical mechanisms) (Trumbore, 2009, Trumbore, 1997).

Assumptions about the existence of a passive pool come from the assumption of inherent chemical resistance of C, radiocarbon dating and the detection of C3-plant residues in SOM of soils that have been cultivated with C4-plants for centuries or longer.

Soil Carbon Initiative | Dow Sustainability Program | Alcoa 9

Figure 2 Conceptual model of SOM dynamics (Figure from Trumbore, 1997, used with permission, Copyright (1997) National Academy of Sciences, U.S.A.). (c) The view on decomposition processes outlined in b) has changed because concepts on the inherent stability of SOM could not be fully substantiated It has been shown that major organic materials (i.e. lignin, cellulose and hemicellulose, lipids and proteins) can decompose fully under ‘perfect’ conditions (i.e. sufficient oxygen supply), even though some organic materials might take longer to decompose than others (i.e.

wood > leaves). In addition to logistical constraints to decomposition, there has been a shift in the science community towards the belief that decomposed plant residues and/or remnants of microbes and fungi and their sorptive interactions with mineral surfaces might be responsible for the stability of parts of SOM.

(d) In the recent literature, the view on the physical nature of SOM shifted from the ‘humic polymer’ model towards a ‘molecular aggregate’ model as proposed by Wershaw (2004) The main assumption of the molecular aggregate model is SOM is composed of partially degraded products of soil biota (i.e. plant polymers). Natural organic matter therefore consists of molecular aggregates (molecular fragments are of amphiphilic nature) of different degradation products that are held together by entropic interactions and/or noncovalent bonds. The model does not provide any structural reason for an inherent stability against decomposition.

Soil Carbon Initiative | Dow Sustainability Program | Alcoa 10

(e) Ultimately, the view of decomposition processes in regards to the term recalcitrance may have changed Recently, the often used concept of ‘recalcitrance’ in soil organic matter research (which has been used as either direct or proxy mechanism within several SOM models) has been pointed out as reification of the concepts of: inherent molecular ‘resistance’ to decomposition; an operationally defined characteristic; and/or a characteristic of long residence time SOM (Kleber, 2010). Kleber (2010) suggests decomposition is perhaps more appropriately viewed as a logistical problem where the key factors are i) the microbial ecology; ii) enzyme kinetics iii) environmental drivers and iv) matrix protection.

This view which constrains SOM decomposition on a logistical basis appears to have greater compatibility to that of a systems approach – that is: structure in terms of organisms (i) and state factors (in terms of enzyme kinetics (ii), environmental drivers (iii) and matrix protection (iv)) can be considered as somewhat determinant of function and likely to contain (potentially) comprehensible feedback mechanisms. These functions could include characteristics such as level of soil aggregation, nutrient cycling, disease suppression etc..

It is currently unclear, how the change of view (in the science community) on the composition of SOM will affect process-oriented models of SOM which were built using the concepts of humification and decomposition (see (b)). Accordingly, should SOC modules be adjusted to new concepts of stabilization as discussed in c, d and e (i.e. as attempted in the Struc-C model by Malamoud et al. (2009), refer to section 4)? Would this adjustment improve the modelling of SOM dynamics? Soil Carbon Initiative | Dow Sustainability Program | Alcoa 11

Soil Carbon Initiative | Dow Sustainability Program | Alcoa 12 3 Soil carbon measurements – Pools and fluxes 3.1 SOC pools Research has indicated that SOM consists of a very complex mixture of (partially) decomposed substances (i.e.

organic molecules such as polysaccharides, lignin, aliphatic biopolymers, tannins, lipids, proteins and aminosugars) which derived from plant litter as well as faunal and microbial biomass (Totsche et al., 2010). Moreover, as discussed it is has been proposed that SOM consists of various functional pools (i.e. Krull et al., 2003, Trumbore, 2009). Subsequently, as demonstrated in Table 1, these pools were divided according to the biological stability of soil organic carbon (labile, stabile, refractory and inert) 2 , the decomposition rate of soil organic carbon (fast-active, slow-intermediate and very slow/passive) and the turnover time of soil organic carbon (short, long, very long).

This ‘division’ led to methods of SOC fractionation.

It is unclear whether new, systems-based models of SOC turnover will be developed in response to new views outlined in section 2.4 and improvements in SOC fractionation methods. Table 1 Fractions of soil organic carbon found in the scientific literature (table adapted from Chapter 2 of the Victorian Government report 2010, based on Baldock (2007)). SOC material Composition Pool category Surface plant residue Plant material residing on the surface of the soil, including leaf litter and crop/ pasture material Fast (or labile) pool Decomposition occurs at a timescale of days to years Buried plant residue Plant material greater than 2 mm in size residing within the soil Fast (or labile) pool Decomposition occurs at a timescale of days to years Particulate organic matter (POC) Semi-decomposed organic material smaller than 2 mm and greater than 50 µm in size Fast (or labile) pool Decomposition occurs at a timescale of days to year ‘Humus’ Well decomposed organic material smaller than 50 µm in size that is associated with soil particles Slow (or stable) pool Decomposition occurs at a timescale of years to decades Resistant organic carbon (ROC) Charcoal or charred materials that results from the burning of organic matter (resistant to biological decomposition) Passive (or recalcitrant) pool Decomposition occurs at a timescale of decades to thousands of years Measurement of SOC The content of SOC has been determined by various techniques.

They can be classified into ex situ methods as well as in situ methods3 (see Chatterjee et al., 2009 for an overview 2 The terms inert and refractory need to be separated. The inert C pool defines material that is ‘biologically not decomposable’ whereas the refractory pool defines material with a long but finite turnover time (see review by Krull et al., 2003).

3 Ex situ methods: wet combustion (oxidization of SOC by an acid solution i.e. Walkley-Black method) and dry combustion (weight-loss-on-ignition such as the muffle furnace, automated CN(S) analysers such as vario EL or vario max) In situ methods: infrared techniques (MIR, NIR), Laser-Induced Breakdown Spectroscopy (LIBS), Inelastic Neutron Scattering, remote sensing

of these techniques). Ex situ methods such as dry combustion are regarded as standard methods to determine the amount of SOC. However, in relation to terrestrial carbon sequestration programs the precise and cost-effective measurement of SOC has become essential in order to verify and potentially monitor SOC stock levels resulting from different management options.

Along with the development of efficient sampling methods on the farmscale level to maintain sampling points at a representative ‘minimum’ (which in turn also lowers the costs involved with sample processing and measurement in the laboratory), in situ analytical methods have also been developed. These methods have the potential of higher cost-effectiveness when compared to traditional methods.

However, for modelling purposes it may be imperative to have well-defined ‘measureable’ carbon ‘fractions’ in order to model SOC dynamics better which has stimulated development of methods of efficient SOM fractionation, ex situ and in situ. SOC fractionation As discussed, SOC has been divided into different fractions based on size, physical protection and (chemical) composition. Various, ex situ often time-consuming, methods have been used to fractionate SOM including physical (density, aggregation) and chemical (solubility, mineralogy) separation procedures (as shown in Table 2). Table 2 Methods of SOC fractionation (i.e.

see review by von Lützow et al., 2007) Physical fractionation Aggregate size fractionation Particle size fractionation Density fractionation heavy fraction (organomineral complexes) light fraction POM (particulate organic matter) macroaggregates (>250µm) /wet sieving /slaking /dispersion (ultrasonic) clay-sized, silt-sized and sand-sized particles /separated using liquids with a certain density (between 1.6- 2.0 g cm-3 ) DOC (dissolved organic matter) /less than 0.45 µm in solution Chemical fractionation Chemical extractions Hydrolysis Oxidation /used to separate ‘humic substances’ into humic acids, fulvic acids and humin /based on solubility in alkali and acid solutions, the most common are NaOH and Na4P2O7 /used to separate hydrolytic bonding of carbohydrates/protein molecules etc.

for instance HF to separate mineral-OM associations /or HCl to quantify proportion of SOC association with proteins, amino acids, amino sugars /used to remove labile/active fraction of SOM (i.e. plant residues) /most common agents are H2O2, NaOCl and KMnO4 Microbial biomass C /i.e. chloroform used as extractant Lately, for the purpose of ‘best partitioning’, integrated fractionation procedures have been applied that include both physical and chemical fractionation methods (i.e. see method proposed by Zimmermann et al., 2007a (Figure 3)). Research is also focusing on separating C pools with the help of cost-effective in situ measuring methods.

For instance, the prediction of Soil Carbon Initiative | Dow Sustainability Program | Alcoa 13

soil C fractions (total organic carbon, particulate organic carbon, resistant-(charcoal) carbon) using mid-infrared spectroscopy (MIR) partial-least-squares (PLS) proved to be promising (Janik et al., 2007, see papers by Zimmermann et al., 2007a). Currently, the ‘Soil Carbon Research Program’ (‘SCRP’ conducted by CSIRO, Australia) is focusing on establishing a spectral library for predicting carbon fractions with MIR techniques. NMR spectroscopy techniques have also been used in the scientific community to investigate the components of SOC. However, they are most likely too cost-intensive for the use of monitoring SOC levels (labile/stable) within an agricultural C trading scheme.

Figure 3 Combined density fractionation procedure (Figure from Zimmermann et al., 2007a, used with permission). Abbreviations: dissolved organic matter (DOC), particulate organic matter (POM), sand and stable aggregates (S+A), silt and clay particles (s+c) and oxidation-resistant carbon (rSOC). Imaging SOC in situ with structure Background One of the biggest challenges in understanding the factors that regulate microbiallymediated processes in soil, including the turnover of carbon, is the complexity of the soil habitat (O'Donnell et al., 2007). Much of what we know about the relation between structure and soil carbon dynamics is obtained using techniques that physically disrupt the soil (Young et al., 2008).

Whilst physical protection has been long cited as a key mechanism in stabilising carbon (Adu and Oades, 1978), there is a lack of useful methods for directly imaging soil structure, carbon and microorganisms. Techniques for creating two-dimensional thin sections of soil exist that preserve microbes in situ and provide some information on the relative Soil Carbon Initiative | Dow Sustainability Program | Alcoa 14

location of microbes and pores (Nunan et al., 2003). However, because of anisotropy and nonstationarity, two-dimensional images on their own provide an incomplete picture. This is especially true in relation to understanding transport processes relevant to resource flows where three-dimensional information on pore topology is required (Werth et al., 2010). 3D imaging of soil X-ray microtomography has been exploited as a tool for imaging soil for several decades and is becoming increasingly accessible as a tool for soil science (Peth et al., 2008). Microtomography has the advantage of providing non-destructive, high-resolution, threedimensional images of soil.

The imaging process essentially returns a three-dimensional map of electron density and the conversion of this into useful information is a crucial first step. Perhaps the most basic distinction to be made is between solid and pore space, but even this can be challenging. Iassonov et al. (2009) demonstrate that vastly different values for the measured porosities result from the application of different algorithms to threshold the solid and void spaces from electron density maps. For this reason it is perhaps not surprising that there has been relatively little progress in mapping the distribution of other components of soil.

Mapping of the three-dimensional distribution of carbon in soil has been limited to 2 studies imaging either freshly applied organic matter in the form of wheat residue (De Gryze et al., 2006) or an artificial system comprising a mixture of sand and particulate organic matter (Sleutel et al., 2008). In both these systems, the challenge of thresholding organic matter from the other phases is reduced by using materials that exaggerate the contrast. De Gryze et al. (2006) were able to study the evolution of structure around the organic matter particles during an incubation experiment and detected significant changes in porosity.

However there remain significant difficulties in identifying soil organic matter even in these systems, with poor signal-to-noise and the lack of standardised thresholding procedures the main impediments (Iassonov et al., 2009, Sleutel et al., 2008). The indirect effects of soil organic matter on the three-dimensional features of the soil microbial habitat have been measured using tomography in a number of studies (i.e. Deurer et al., 2009, Feeney et al., 2006, Papadopoulos et al., 2009). In Papadopoulos et al (2009) and Duerer et al. (2009), the studies focus on the difference between organic and conventional production systems.

In contrast, Feeney et al. (2006) compare soil in the vicinity of roots to soil that is more distant and soil without roots. In each case it has been observed that SOM has a profound impact on the physical structure of soil resulting in increases in porosity. Imaging of soil water has been achieved both with (Carminati et al., 2008) and without (Tippkotter et al., 2009) contrast agent being added to the water. Carminati et al. (2008) were able to image the change in the distribution of water at the contact between two soil aggregates at successive water potentials at a resolution of 6μm. Tippkotter et al.

(2009) imaged the distribution of water films in macropores in undisturbed soil. While confirming that pores

(up to 30μm) lining the larger pores. The nature and authenticity of these thick-films is the subject of further investigation, but the results clearly illustrate the potential power of tomography to reveal new features of the soil habitat. As the technology advances, the possibilities increase and there have been attempts to image microbial cells using x-ray microtomography. Le Gross et al. (2005) report on the use of x-ray microscopy to image intact bacterial cells at a resolution of up to 30nm. Thieme et al. (2000) use tomographic analysis and x-ray microscopy to image bacterial cells in flocs of colloidal soil particles at a resolution of 100nm.

The technique involves fast-freezing of the sample and because of the high resolution the sample size is limited (6 µm in this case). Therefore the scope for extending this approach to larger soil features may be limited at present.

Whilst simple analyses of tomographic images is revealing, the most significant advances are to be made in the study of the consequences of the observed structure for function. For this we need more sophisticated modelling approaches and several promising methods have been developed that can help link the observed structures to the consequences for the microbial community. Particle-based simulation of flow including the lattice-Boltzmann method can be used in conjunction with microbial growth models to study the interaction between structure and microbial processes in porous media (von der Schulenburg et al., 2009).

In this case, the growth of the biofilm changes the flow by restricting parts of the pore space and the evolution of the conductivity over time can be simulated. A comparison of twoand three-dimensional simulations show qualitative differences and highlights the importance of full three-dimensional characterisation of the porous structure. A model for soil organic matter decomposition in structured soil has been developed by Monga et al. (2009). In contrast with von der Schulenburg et al. (2009), the structure does not affect the transport of resources, but influences the relative proximity of organic matter and microorganisms in the soil.

It also affects the distribution of soil water, which is assumed to act as a bridge connecting resources to microbes. Therefore the effects of soil structure on the co-location of microbes and resource can be studied by simulating the effects of varying the water content. Clearly this opens up opportunities in the future for rigorous testing of the model.

To date, models for the interaction between structure and microbial activity assume that the microbes are bacteria. However, it has been known for some time that fungi play a crucial role in carbon dynamics (Strong et al., 2004, Young et al., 2008). To date there is only one model that is capable of studying the effects of soil structure on fungal growth (Falconer et al., 2008). Furthermore, in all models so far, the structure is assumed to be static. This is at odds with the generally accepted conceptual model for soil aggregation (Oades, 1993, Tisdall and Oades, 1982). The dynamics of structure is clearly important in the context of physical protection of organic matter in soil.

Furthermore, the fact that microbial activity is coupled to structure generation which in turn affects resource availability could have profound affects for the dynamics of the system (Young and Crawford, 2004). These are potentially important omissions from our understanding of the role of the microbial habitat in carbon turnover. Soil Carbon Initiative | Dow Sustainability Program | Alcoa 16

3.2 C fluxes Fluxes of carbon entering soils are directly linked with both, the soil carbon pools and the soil carbon effluxes (SCE). Root exudates fuelled by photosynthesis and decomposition of aboveground and belowground litter are carbon suppliers to soils. Along with decomposition of native soil organic matter and root respiration these factors are also the major control of soil respiration or SCE (Luyssaert et al., 2007). Soil respiration is defined as carbon dioxide (CO2) released from soil to the atmosphere via the combined activity of (1) roots (root respiration) and associated microorganisms, mainly respiring the recently-assimilated C by plants (autotrophic respiration), and (2) microand macroorganisms decomposing litter and organic matter in soil, referred to as heterotrophic respiration (Högberg et al., 2005).

Because root exudates are mostly highly labile whereas leaf and root litter are rather less labile and native, soil organic matter is usually the least labile. Each component returns C to the atmosphere on different time scales: rapid C cycling is associated with root/rhizosphere respiration (days to months), followed by litter C decomposition (years), and then inherent (native) soil organic matter decomposition (decades to centuries) (Cisneros-Dozal et al., 2006, Kuzyakov and Larionova, 2005). Thus, thanks to the relatively slow turnover rate of soil organic matter, heterotrophic respiration has profound implications for long-term storage of organic C in soil.

Measurement of C fluxes The stable isotope δ13 C of carbon is used in different ways to study the dynamics of soil carbon: Pulse-labelling is used to trace the flux of recently-fixed C through the plant-microbesoil continuum (Hogberg et al., 2008), while isotopic analysis of SOM from disturbed ecosystems that experienced a shift from a dominance of C3 plants to a dominance of C4 plants (e.g. tropical forests to crops or grasslands) gives information on soil carbon cycling over the years or decades following land use changes (Trumbore, 2000, Ehleringer et al., 2000).

Radiocarbon 14 C is also used to estimate the origin of the carbon in both soil pools and in soil effluxes (Trumbore, 2000, Trumbore, 2009).

Belowground carbon allocation can be estimated with a carbon balance approach equal to soil respiration minus carbon inputs from aboveground litter plus changes in carbon stored in roots, litter and soil (Giardina and Ryan, 2002). Soil respiration is usually monitored in situ using dynamic or static chamber methodologies coupled with an Infra Red Gas analyser (IRGA) to measure CO2 efflux at the soil surface (Luo and Zhou 2006). It can be further refined by separating components of soil respiration (Hanson et al., 2000, Högberg et al., 2005, Kuzyakov, 2006) using methods grouped into four broad categories: (1) integration of physically-disintegrated respiratory components (root, leaf litter, soil organic matter, etc); (2) exclusion of live roots from soil monoliths via trenching; (3) the use of non-invasive stable or radioactive isotopes that rely on different ‘isotopic signatures’ of CO2 derived from living roots and soil organic matter; and, Soil Carbon Initiative | Dow Sustainability Program | Alcoa 17

(4) stem girdling in forest ecosystems (involving instantaneous termination of photosynthetic C flow to roots and associated microorganisms without affecting micro-climate, at least initially (Högberg et al., 2005). Soil Carbon Initiative | Dow Sustainability Program | Alcoa 18

Soil Carbon Initiative | Dow Sustainability Program | Alcoa 19 4 Soil carbon modelling 4.1 Process oriented versus organism-oriented models Model characteristics As discussed, relevant changes in land use and land management practices have the opportunity to sequester soil C (i.e.

review by Guo and Gifford (2002)). It is therefore imperative to assess soil organic matter dynamics and the sequestration potential of soil C (preferably) at the landscape scale (Post et al., 2007) to simulate the response of the soil system to environmental pedoturbations (Smith et al., 1998). Models of soil organic matter dynamics are a well-known tool to address these demands (i.e. they give farmers an indication of how much SOM to expect when applying certain management practices4 ). Ultimately, they should provide us with an indication about the size of the C stock under different soil types, management practices and climate regimes.

As noted by Batlle-Aguilar et al. (2008) and Post et al. (2007) they can be divided into four categories depending on their internal structure (1) process-oriented, (multi)-compartment models (2) organism-oriented (food-web) models (3) cohort models describing decomposition as a continuum and (4) a combination of model types (1) and (2)5 .

In the following section process-oriented and organism-oriented models and their benefits and limitations will be discussed in more detail (see review by Manzoni and Porporato (2009) for a list of main features of SOM dynamic models, Table A1 and A2). Table 3 gives an overview of the main characteristics of these models. 4 See Chapter 4 of the Victorian Government report 2010. 5 Combined models of processes-oriented and organism-oriented models are often limited by the data required to describe the organism-oriented module. Cohort models separate the soil organic matter into different cohorts which share specific characteristics.

These cohorts are then further divided into different pools (i.e. carbon and nitrogen). Rather than simulating decomposition rates based on physical and biochemical processes, as performed in process-oriented models, decomposition is simulated based on the physiology of the microbial biomass. SOMCO (soil organic matter cohort by Gignoux et al. 2001) is cited as one example of a cohort model in the scientific literature. Here, cohorts are divided depending on similar ages and new cohorts are created after each time step following existing cohorts. Cohorts continue in this network until their relative amount of soil organic matter becomes negligible (see review by Battle-Aguilar et al., 2010).

Table 3 Main characteristics of processes-oriented versus organism-oriented models (based on (Lindsay, 2008, Brussaard, 1998, Krull et al., 2003, Post et al., 2007, Smith et al., 1998, Dokuchaev, 1899)) Process-oriented models Organism-oriented models Model type mechanistic predictive value mechanistic explanatory value Aim /simulate processes involved in SOM migration and transformation /simulate SOM using functional/taxonomic groups of the soil Examples CANDY, CENTURY, DAISY, DNDC, ITE, NCSOIL, RothC, Socrates, SOILN, SOMM, Struc-C and the Verbene model /Fungal-growth models /Models of decomposition of OM that incorporate functional groups of microbial biomass /Food web models based on taxonomic groups (mostly detrital models) Representation of SOM /different conceptual carbon pools with similar chemical or physical characteristics /differ by decomposition rates, stabilization mechanisms /generally soil biota only included in form of microbial biomass (exception: SOMM) Generally, more than one compartment of SOM degradation: a) active pool (fresh plant material, root exudates, microbial biomass) with MRT of 1 year b) slow pool (SOC that decomposes at intermediate rate) with MRT of 100 years c) passive or inert pool (SOC with physical or chemical stability) with MRT of 1000 years SOC dynamics represented through different pools of soil biota (classified according to their taxonomy or metabolism) i.e.

representation of soil biota by functional groups (food web models): microorganisms (bacteria, mycorrhizal and saprotrophic fungi) SOM and litter (represented in form of roots, detritus) Mechanism /SOC decomposition based on first-order kinetic rates /C and N fluxes simulated through functional groups based on their specific death rates and consumption rates, applying energy conversion efficiencies and C:N ratios of the organisms Time-step /weekly or monthly /daily Scale /include top 30 cm of the soil /small-plot to regional-scale /small-plot Application /have been applied to a range of ecosystems (grassland, arable land, grass-arable rotations, forest) /have been applied to arable land and grassland Others /successfully coupled with GIS software (CANDY, CENTURY, RothC) /include changes of soil biota communities in the modelling of SOM dynamics (i.e.

simulating feedback mechanisms due to changes in biota activity or characteristics) Process-oriented models The main characteristics of the most popular process-oriented soil organic carbon and nitrogen turnover models have been discussed frequently in the literature (Table 4). Their representation of SOM dynamics is displayed in Figure 4 to 5. Accordingly, the CENTURY, RothC and DNDC model are the most frequently used models to simulate SOM dynamic spatially on the small farm-scale level (i.e. Viaud et al., 2010). Soil Carbon Initiative | Dow Sustainability Program | Alcoa 20

Table 4 Main characteristics of the most frequent referred to models in the scientific literature (based on Smith et al. 1997, Grace et al. 2006, Krull et al. 2003) Model Main characteristics Reference CANDY /modular system combined with data base system for model parameters, measurement values, initial values, weather data, soil management data /simulates soil N, temperature and water to predict N uptake, leaching, water quality /uses proportion of soil particles to separate IOM (

process rates regulated by N and ash content of litter fall /3 soil litter layers: L, F, H /soil animals influence C fluxes (i.e.

distinction into forms of humus such as mull and mor based on role of soil fauna – microarthropods and earthworms) /models C accumulation in soil organic horizons Komarov, 1996) Struc-C /updated, modified version of the Roth-C model /monthly time step /incorporates soil structure (aggregate) hierarchies within physical protection of SOC /simulates formation of organo-mineral associations and aggregates (physically protected SOC) (Malamoud et al., 2009) Verbene /developed for grasslands /implements a plant growth submodel /3 submodels: soil water, SOM (plant residues: decomposable, structural, resistant, OM: stabilized, protected, unprotected), soil N / physical protection caused by soil clay /decomposition rate modified by temperature and soil moisture, not influenced by microbial activity /biomass included (i.e.

Verbene et al., 1990) Figure 4 Structure of the Rothamsted Carbon Model (based on Cleman and Jenkinson, RothC-26.3, Model description and users guide).

Soil Carbon Initiative | Dow Sustainability Program | Alcoa 22

Figure 5 Structure of the CENTURY model (based on CENTURY Model Version 5, User’s guide). Apart from CANDY and ITE, the majority of these models incorporate the biomass of the soil in at least one compartment. However, only SOMM describes the mesoand macrofauna and distinguishes different forms of organic matter based on the abundance of the soil fauna (Smith et al., 1998). Smith et al. (1899) tested the performance of nine SOM models on predicting long-term changes in SOM using data from seven long-term (>20 years) experiments.

In summary, the performance of most processes-oriented models showed a high applicability for predicting long-term soil organic matter dynamics (decades) across a range of land uses, soil types and climatic regions (Smith et al., 1998). However, two groups of models were distinguished in relation to their overall performance and similarity in modelling errors: (1) RothC, CANDY, DNDC, CENTURY, DAISY and NCSOIL and (2) SOMM, ITE and Verbene. When compared to each other, the second group of models showed significantly larger model errors than the models in the first group. The coupling of SOM modules with more defined physiologically based plant growth models (such as in ITE) was named as one of the main reasons for their relatively poor performance.

Smith et al. (1997) postulated that too much detail most likely introduced a greater degree of error and uncertainty when applied to a range of land uses since models whose SOM modules were coupled with simpler plant Soil Carbon Initiative | Dow Sustainability Program | Alcoa 23

You can also read
Next part ... Cancel