The soil carbon cycle#
Soils are one of the largest stores ecological stores of carbon. For this reason, there has been long term interest in modelling the soil carbon cycle. This is a massively complex task as carbon exists in the soil in a huge variety of forms, therefore soil carbon modelling (starting with the CENTURY model (Parton et al., 1987)) have generally grouped carbon into a small set of pools with common properties. This is the broad approach that we take in the Virtual Ecosystem. In this page, we set out the set of carbon pools that our soil model uses, explain the inputs that these pools receive from the other modules of the Virtual Ecosystem, and describe the processes that transfer carbon between pools.
Soil carbon pools#
Historically, the predominant framework for modelling soil carbon has been the CENTURY model (Parton et al., 1987), which divides soil organic matter into three pools (active, slow and passive). These pools are characterised primarily by their turnover rates, but are also differentiated by lignin content of the organic matter that flows into each pool. This framework has come under sustained criticism as these pools are conceptual and not directly measurable. In response to this there has been a movement towards using soil carbon pool definitions that are based upon measurable physical and chemical properties. The Millennial model combines the most commonly used of these soil carbon pools into a single model (Abramoff et al., 2018). This model framework is both comprehensive and defines measurable pools, and for this reason we make use of a variant of it in our soil model.
Where we differ from the Millennial model is that we only include pools that represent the chemical protection of carbon, and neglect the pool that represents the physical protection of carbon (i.e. soil aggregates). This partly to avoid double counting of protection mechanisms, as it is quite common for carbon that is protected chemically to also be protected physically. Additionally, pools are appropriately set up to represent chemical transformations (through enzymatic kinetics), but are less appropriate for physical transformations (i.e. soil aggregates changing in size is hard to capture with a discrete set of pools). Properly capturing physical protection of carbon requires the distribution of particle sizes in soil, which is not something we plan to add in the immediate future.
The carbon pools that we use in our soil model are as follows:
Particulate organic matter (POM)#
Particulate organic matter (POM) derives from the decomposition and fragmentation of litter and other necromass. It can be formed from plant material, insect carcasses, aggregates, fungal matter, etc. Generally, the particulates are of sufficient size that their original source can still be identified. In most systems this is a pool with a reasonably fast turnover rate (order of months). However, in heavily waterlogged soils (i.e. peatlands) this pool turns over far more slowly and is a significant store of carbon.
Low molecular weight carbon (LMWC)#
Low molecular weight carbon (LMWC) consists of molecules that are simple, soluble and labile, i.e. those that are immediately utilisable by microbes. It is formed through the microbially mediated breakdown of more complex carbon, but is also directly supplied by plant roots. LMWC is commonly lost to leaching or microbial uptake. This pool turns over rapidly (order of days).
Mineral associated organic matter (MAOM)#
Carbon can be protected from microbial activity by mineral association, whereby mineral surfaces take up organic matter by adsorption, conferring chemical protection. This pool turns over very slowly (order of years to decades) and so in most soils it is the main form of (chemically) protected carbon.
Microbial biomass#
Microbial biomass accounts for a small fraction of total soil carbon. This pool turns over rapidly (order of days) and only represents a very small fraction of total soil carbon. However, microbes are key drivers of soil carbon cycling. It is therefore very important to track the size of the microbial biomass pool because a significant amount of carbon flows through it, with microbial respiration being one of the major sources of carbon loss from the system.
Microbial necromass#
When microbial cells die they break down and form the microbial necromass. This consists of complex biochemicals that normally would be contained within cells, but are now exposed directly to the soil environment. This pools turns over rapidly (order of days), and is very small. However, it is important to track this pool as the biochemicals that it represents rapidly bind to soil minerals, so the size of this pool can affect how quickly new protected carbon is formed.
Soil carbon inputs#
Plant inputs#
Most plant inputs enter to the soil via the litter model, due to litter mineralisation. A portion of this is assumed to have occurred due leaching of simple compounds from the litter into the soil. This part of the litter mineralisation flux gets added to the LMWC pool, and is calculated by
where \(C_l\) is the fraction of litter carbon decomposition that happens by leaching and \(M_C\) is the total rate of carbon mineralisation from the litter. The remainder of the litter mineralisation is assumed to be in a more complex form and gets added to the POM pool with rate
Plants also directly provide carbon to the soil in the form of root exudates. These root exudates are simple carbohydrates, so this input flux is added to the LMWC pool.
Animal inputs#
The animal model contains excrement and carcass pools which are available to scavengers. A certain portion of these pools is assumed to periodically decay into the soil. As breakdown is already implicitly modelled within the animal model, we assume that these animal inputs to the soil go solely into the LMWC pool.
Exchanges between soil pools#
Microbial uptake and growth#
Microbes take up LMWC both as a source of energy and as a source of carbon. These resources are then used to synthesis new biomass, replace cells that have died and proteins that have degraded, and to produce extra-cellular enzymes. The net change in the size of the microbial pool is given by
where \(\lambda\) is the rate of new biomass synthesis, \(d\) is the rate at which biomass is lost to cell death and protein degradation, and \(P_E\) is the rate at which enzymes are produced.
Enzyme mediated decomposition#
Both POM and MAOM are broken down into LMWC by enzyme-mediated reactions. As mentioned above, these enzymes are produced by microbes, and are pool specific (i.e. one enzyme class breaks down POM and the other breaks down MAOM). The rate of these decomposition processes is given by
where \(P_i\) is the concentration of the resource type \(i\), \(E_i\) is the concentration of the relevant enzyme class, \(k_i\) is the decomposition rate constant, \(K_i\) is the decomposition saturation constant, \(f_{T,r}\) is a factor capturing the impact of temperature on the process rate, \(f_{T,s}\) is a factor capturing the impact of temperature on the concentration at which the enzyme saturates, \(f_W\) is a factor capturing the impact of soil moisture on the process rate, \(f_{p}\) is a factor capturing the impact of soil pH on the process rate, and \(f_{c}\) is a factor capturing the impact of soil clay content on the concentration at which the enzyme saturates. See the definitions of environmental factors for more detail.
Microbial turnover#
The rate at which microbial biomass is lost to both cell death and protein degradation (\(d\)) is temperature dependent. All of these losses get added to the necromass pool. The breakdown of this necromass pool to form LMWC is modelled using linear kinetics as
where \(k_d\) is the rate constant for necromass breakdown and \(N\) is the size of the necromass pool. The sorption of necromass to soil minerals to form MAOM is also modelled using linear kinetics as
where \(k_s\) is the rate constant for necromass sorption. The ratio of \(k_d\) and \(k_s\) will be the same as the ratio between the amount necromass that becomes LMWC and the amount that becomes MAOM.
Mineral sorption and desorption#
MAOM can also be formed via sorption of LMWC. The carbon associated with the soil surface can also desorb, this leads to a decrease in the size of the MAOM pool and a corresponding production of LMWC. We model both of these processes using linear kinetics, the resulting net change in the size of the LMWC pool can be expressed as
where \(K_d\) is the rate constant for MAOM desorption, \(K_s\) is the rate constant for LMWC sorption, \(M\) is the size of the MAOM pool and \(L\) is the size of the LMWC pool. Most MAOM formation occurs via necromass sorption, therefore the default value for \(K_s\) is small relative to \(k_s\).
Removal of soil carbon by water#
Removal of nutrients from the soil by water occurs when water moving out of the microbially active region of the soil carries dissolved nutrients away with it. By definition, any organic matter that is simple enough to solubilise is included in the LMWC pool, so this is the only soil carbon pool to be affected by water flows. The functional form that we assume nutrient removal by water follows is provided in the soil-abiotic environment links documentation.