« Previous | Contents | Next »
Listen
Appendix 2 Development and description of the ECOSSE model
A2.1 Modularisation of SUNDIAL/ MAGEC
Figure A2.1 Modularisation of SUNDIAL- MAGEC
The model was restructured into modules that contain routines describing soil C and N turnover (initialisation, additions, microbial and physical processes), crop processes (growth, N demand and return of C and N in debris), and the soil water processes (drainage and evapotranspiration). This allows easy modification of individual processes, replacement of modules with new approaches, and independent evaluation of individual modules (Figure A2.1).
A2.2 Improved description of N 2O production
Figure A2.2 Emissions of N2O
In SUNDIAL- MAGEC, N loss by denitrification is modelled using a simple linear relationship. Losses are not partitioned into N 2, N 2O and NO gas, so emissions of greenhouse gases cannot be estimated. In ECOSSE, emissions of N are simulated due to denitrification and partial nitrification (Figure A2.2). The emissions are then partitioned into emissions of N 2, N 2O and NO.
Denitrification: Denitrification is a process that responds to changes in the system on a shorter timescale than many of the other C and N turnover processes. As a result, denitrification is difficult to measure accurately, and difficult to simulate without short-term and detailed input data. As a result, many models of denitrification have high data requirements. The aim in ECOSSE is to develop a model that will simulate processes to the required accuracy but without the need for detailed data that will not be available at the large scale.
The denitrification sub-model of DNDC calculates N 2O and NO production, consumption and diffusion during rainfall, irrigation and flooding events. DNDC simulates relative growth rates of nitrate, nitrite, NO, and N 2O denitrifiers based on soil heterotrophic respiration, and concentration of DOC and N oxides. The soil matrix is divided into aerobic and anaerobic zones using an "anaerobic balloon", which swells and shrinks in response to oxygen diffusion and consumption in the soil profile. Substrates allocated to the anaerobic zone of the profile are then used to determine the denitrifier growth rates based on a simple function of multinutrient-dependent growth (based on Bader, 1978). The death rate is a constant fraction of total denitrifier biomass. Relative growth rates for denitrifiers with different substrates are assumed to be independent, and competition takes place via the common DOC substrate (based on Leffelaar & Wessel, 1988). Substrate consumption rates are calculated using growth rates and biomass. NO, N 2O and N 2 fluxes are calculated using the basic laws of kinetics, as denitrification is a typical sequential reaction. As intermediates of the reactions, NO and N 2O fluxes are determined by rates of production, consumption and escape from the reacting system. Escape is controlled by diffusion rate, a function of soil porosity, moisture, temperature and clay content (Li, 2000).
The CENTURY denitrification model in NGAS is based on data from Weier et al. (1993). Total denitrification N gas fluxes (N 2 + N 2O) are a function of soil heterotrophic respiration rate (index of carbon availability), soil NO 3- levels, and soil water filled pore space. Soil NO 3- levels and respiration rate determine the maximum total N gas flux according to s-shaped curves fitted to experimental data. This maximum is then reduced by a fraction determined by water filled pore space, as a function of soil texture. Maximum denitrification occurs where water filled pore space is 0.9 or above. Below this value, denitrification is reduced exponentially, with fine soils declining most rapidly, followed by medium and then coarse soils. Denitrification drops to zero at water filled pore space of 0.6 for fine soils, 0.5 for medium and 0.4 for coarse. Total denitrification N gas flux is then partitioned into N 2 and N 2O as a function of soil NO 3-, WFPS and soil respiration (Parton et al., 1996).
In ECOSSE, a simpler approach is used in which total denitrification is simulated as a proportion of the nitrate content of the layer, modified according to water content and biological activity of the soil (as described by CO 2 release during decomposition). The denitrified N is then divided into N 2, N 2O and NO according to the water and nitrate content of the soil The total loss of N due to denitrification is given by
,
where N d is the amount of nitrogen emitted from the soil during denitrification (kg N ha -1); m NO3 modifies this amount depending on the nitrate level, m water depending on the water content, and m bio depending on the biological activity of the soil; and NO 3 is the amount of nitrate in the soil (kg N ha -1).
The nitrate modifier, m NO3, is based on the model developed by Henault & Germon (2000), i.e.
.
The response of the denitrification process to the amount of nitrate in the soil is shown in Figure A2.3 (A).
The water modifier, m water, is based on the fitted equation of Grundmann & Rolston (1987), i.e.
where W is the water content and W max is the maximum water content of the soil (mm/layer). The response of the denitrification process to the soil water content is shown in Figure A2.3 (B).
The biological activity modifier, m bio, is based on the relationship developed by Bradbury et al. (1987), i.e.
where CO 2 is the amount of CO 2 produced during mineralization (kg C ha -1). This acts as a surrogate measure of the soil biological activity. The response of the denitrification process to the soil biological activity is shown in Figure A2.3 (C).
Figure A2.3 Response in amount of N emitted due to denitrification to (A) nitrate in soil; (B) soil water content; (C) biological activity
The N lost by denitrification, N d, is partitioned into N 2 and N 2O using simple linear relationships shown in Figure A2.4.
Figure A2.4 Partitioning of N lost by denitrification into N2 and N2O with respect to soil water and nitrate content
The amount of N 2 gas lost by denitrification, N d,N2, is given (in kg N ha -1) by
,
where is amount of N 2 gas lost by denitrification (kg N ha -1), p water proportions denitrification into N 2 according to the water content of the soil, 
where d N2,FC = 0.5 is the proportion of denitrified N lost as N 2 at field capacity), and p NO3 proportions denitrification into N 2 according to the nitrate content of the soil, 
where d zero = 0.1 is the proportion 
at which N 2 emission falls to zero and all denitrified N is lost as nitrate.
The amount of N 2O gas lost by denitrification, N d,N2O is similarly calculated (in kg N ha -1) by the following expression,
.
Nitrification:DNDC predicts nitrification rates by tracking nitrifier activity and NH 4+ concentration. Growth and death rates of nitrifying bacteria are calculated as a function of DOC concentration, temperature and moisture, based on Blagodatsky & Richter (1998) and Blagodatsky et al. (1998). Nitrification rates are then predicted as a function of the nitrifier biomass, NH 4+ concentration and pH. Nitrification-induced NO and N 2O production are calculated as a function of nitrification rate and temperature (Li, 2000).
NGAS takes a much simpler approach, calculating nitrification directly as a function of water filled pore space ( WFPS), pH, temperature and soil NH 4+ levels. Each of the four variables controls the fraction of nitrification that occurs under the given conditions. The effect of WFPS is based on Doran et al. (1988) and is a function of soil texture. It consists of two bell-shaped curves where maximum nitrification occurs at of 0.55 for the sandy soils and 0.61 for medium and fine-textured soils. The effect of temperature is an exponential function with a Q 10 of 2, based on the work of Sabey et al. (1959):
where T /eff describes the effect of temperature on nitrification, and T is the air temperature at the soil surface (ºC).
The effect of pH on nitrification is an S-shaped curve based on data presented by Gilmore (1984) and Motavalli et al. (1996), where the optimum pH is taken to be pH 7. Finally, NH 4+ affects nitrification based on Malhi & McGill (1982):
where NH4 eff is the effect of NH 4+ on the nitrification fraction and NH4 is the concentration of ammonium in the soil.
Maximum nitrification is set as a constant based on field data from Mosier et al. (1991), although (Parton et al. 1996) suggests it varies as a function of soil texture. None of these factors are parameterised for highly organic or very acidic soils. Frolking et al. (1998) modified these relationships when incorporating NGAS into CENTURY, changing the temperature curve according to Malhi & McGill (1982), specifying a maximum nitrification rate of 10 % of soil NH 4+ per day, and setting N 2O emissions as 2 % of nitrification.
Using a similar but simpler approach, nitrification is simulated in ECOSSE according to the amount of ammonium in the soil layer, and is modified according to the temperature, moisture content and soil pH. The amount of N nitrified, N n, is calculated (in kg N ha -1) using the expression for nitrification developed by Bradbury et al. (1987), i.e.
,
where NH 4 is the amount of ammonium in the soil (kg N ha -1), m_ temp is a rate modifier due to soil temperature, m_ water is a rate modifier due to soil water, and m_ pH is a rate modifier due to soil pH.
The expression for the water rate modifier is adapted from that used by Bradbury et al. (1987) to include reduced nitrification in anaerobic soils, i.e.
,
where d (-1bar) is the water deficit on the soil layer at -1bar (mm), and m_ anaer reduces nitrification in anaerobic soils. If the W < 0.9 W max, m_ anaer = 1, otherwise m_ anaer is given by 
.
The response of nitrification to the soil water content is shown in Figure A2.5 (A).
The expression for the temperature rate modifier, m_ temp, is given by the expression used by Bradbury et al. (1987) for both mineralization and nitrification, i.e.
,
where T is the air temperature (°C).
The response to air temperature is shown in Figure A2.5 (B), and the response to ammonium is shown in Figure A2.5 (C).
The pH response function suggested by Parton et al. (1996) was used. In this model, rate is about 1 (unmodified) at about neutral pH, but decreases below neutral:
Figure A2.5 Response in amount of N nitrified to (A) soil water content; (B) temperature; (C) ammonium content
.
Conversion of ammonium to nitrate by nitrification is accompanied by gaseous losses of N due to complete and partial nitrification. The gaseous losses as NO and N 2O are calculated using linear relationships shown in Figure A2.6.
Figure A2.6 Partitioning of N lost by denitrification into N2 and N2O with respect to soil water and nitrate content
The amount of gas emitted as N 2O during nitrification, N n,N2O, is given (in kg N ha -1) by
,
where n FC = 0.2 is the proportion of N 2O produced due to partial nitrification at field capacity, n gas = 0.02 is the proportion of full nitrification lost as gas, and n NO = 0.1 is the proportion of full nitrification gaseous loss that is NO.
Similarly, the amount of gas emitted as NO during nitrification, N n,NO is given (in kg N ha -1) by
.
A2.3 Description of methane production and oxidation
Methane is an important contributor to global warming, which is produced by methanogenic bacteria in soil when decomposition occurs under anaerobic, reducing conditions. Wetlands represent the most important natural source of methane emissions to the environment. As the rate of methane emission is often reported to increase with temperature, there is potential for a positive feedback due to climate change. This emphasises the need to understand the processes that control methane emission from wetlands and how they react to both environmental and land use changes.
DNDC calculates CH 4 production as a function of DOC concentration and temperature, under anaerobic conditions where soil Eh is predicted to be 150 mV or less. CH 4 oxidation is calculated as a function of soil CH 4 and Eh. CH 4 moves from anaerobic production zones to aerobic oxidation zones via diffusion, which is modelled using concentration gradients between soil layers, temperature and soil porosity. CH 4 flux via plant transport is a function of CH 4 concentration and plant aerenchyma. Plant aerenchyma is a function of the plant growth index, which is calculated using plant age and season days. If plant aerenchyma is not well developed, or soil is unvegetated, the efflux is determined by ebullition (bubbling). In DNDC, this is assumed to occur only at the surface level, and is regulated by CH 4 concentration, temperature, porosity, and any existing plant aerenchyma.
By contrast, to the very complex approach used in DNDC, Christensen et al. (1996) describe methane emissions very simply as a proportion of the total heterotrophic respiration. In ECOSSE we simulate methane emissions using a process-based but simple approach, as the difference between methane production and methane oxidation, the oxidation process adding to emissions of carbon dioxide (Figure A2.7).
Figure A2.7 Structure of methane model in ECOSSE
Methane production during anaerobic decomposition is simulated using a similar pool approach as is used for aerobic decomposition. The amount of anaerobic decomposition ( D an) is calculated using the following expression:
,
where C pool is the amount of carbon in the biomass, humus, decomposable plant material or resistant plant material pool; k pool is the rate constant for decomposition (initially assumed be the same as in aerobic decomposition: for decomposable plant material k DPM = 10 yr -1; for resistant plant material k RPM = 0.3 yr -1; for soil biomass k BIO = 0.66 yr -1; and for humus k HUM = 0.02 yr -1); m ///water is the rate modifier for moisture; m ///temp is the rate modifier for soil temperature; and m ///pH is the rate modifier for pH. The difference between the rates of aerobic and anaerobic decomposition is simulated through the different functions describing these rate modifiers. Derivation of the rate modifying factors is described in the following sections.
The production of methane 
is then given by
where a is the proportion of decomposing materials partitioned to biomass, and b is the proportion partitioned to humus. The values of a and b are calculated from the efficiency of decomposition 
and 
. The efficiency of decomposition is set for each given soil type, and is currently assumed to be the same as for aerobic decomposition.
The oxidation of methane 
is calculated from methane production as
where t is a transport factor (for non-transporters t = 0; for transporting non-sedges t = 0.25; for transporting sedges t = 1 (Kettunen, 2002); n is a soil dependent factor that accounts for different rates of diffusion and oxidation (derivation is described in the following sections); and d is the depth (cm).
Soil Temperature Rate Modifier, m ///temp
Field measurements of changes in methane emissions with temperature can be difficult to unravel, as many confounding factors can contribute to the observed emissions. Christensen et al. (2002) report that mean seasonal temperature is the best predictor of methane fluxes on a large scale. Hargreaves et al. (2001) observed an exponential relationship between surface temperature (0-10 cm) and methane flux for measurements without a thaw period, with a Q 10 of 4, while Rask et al. (2002) found a linear relationship between the same factors in a shallow bay area of a fen. Hargreaves & Fowler (1998) reported a linear relationship with temperature between 7 and 11°C, with a slope of 5 _mol CH 4 m -2 h -1 °C -1 for peat wetlands in Caithness, Scotland. Other workers have also reported significant relationships between soil temperature in the surface layer and methane flux, in wet tundra sites (Christensen et al. 1995) and subalpine wetlands (Wickland et al. 1999, 2001). Although these field observations describe different responses, a positive relationship between methane emissions and temperature is generally observed (Figure A2.8).
Figure A2.8 Reported responses of methane emissions to temperature
Micro- and mesocosm measurements provide a less complex picture, allowing confounding factors to be removed from the experimental setup. Daulat & Clymo (1998) reported an exponential relationship between soil temperature at 5 cm depth and mean methane flux in peat cores from Scotland. Lloyd et al. (1998) commented on the high sensitivity of methane fluxes from Scottish peat cores to temperature, reporting a Q 10 of 3 in the dark. MacDonald et al. (1998) also report Q 10 values of around 3 (between 5 and 15ºC) for relationships between peat temperature and methane flux from peat cores from Northern Scotland, which are linear under semi-natural conditions, and exponential under controlled conditions of constant humidity and light.
Following the model of Kettunen (2002), the temperature rate modifier is given by the equation:
where T soil is the temperature of the soil layer (ºC)
In agreement with the observations of Daulat & Clymo (1998), Lloyd et al. (1998) and MacDonald et al. (1998), this relationship shows a Q 10 value close to 3 between 5 and 15°C (see Figure A2.9).
Figure A2.9 Response of anaerobic decomposition to soil temperature
Soil Water Rate Modifier, m ///water
Methane emissions only occur in strongly anoxic soils (Le Mer & Roger, 2001), therefore the rate modifier is assumed to be non-zero only at water contents over field capacity.
Following the approach used wetlands DNDC (Zhang et al., 2002), the rate of methane production is assumed to increase exponentially above field capacity (see Figure A2.10) and is calculated as
where 
is the available water in a 5cm layer (mm); and c 1 and c 2 are fitted constants ( c 1=0.5 and 
where 
is the available water at saturation (mm)).
Figure A2.10 Response of anaerobic decomposition to available water in the soil
Soil pH Rate Modifier, m ///pH
Methanogenic bacteria are generally reported to exhibit maximum activity under neutral or slightly higher pH conditions (Garcia et al. 2000) and to be very sensitive to variations in soil pH (Wang et al., 1993). Garcia et al. (2000) reported that 68 species of methanogenic bacteria could not grow at a pH lower than 5.6.
However, methane producers can adapt to more acidic environments, as many studies have recorded methanogenic activity in soils with a lower pH. Williams and Crawford (1985) reported that a mixed bacterial culture from a Minnesota peatland produced methane at pH values between 3 and 4. Dunfield et al. (1993) investigated methane production in peat soil samples from temperate and subarctic areas (pH 3.5-6.3) and reported an optimum production rates at pH of 5.5 to 7.0. Inubushi et al. (2005) reported a positive correlation between methane production activity and soil pH (r 2 = 0.802, P <0.01) for peat soil samples from a temperate Japanese wetland, which had a pH range of 5-7. However, more acidic Indonesian peat soils, which ranged from pH 3.9-5, showed no correlation with pH (Inubushi et al. 2005).
Depth can also affect the influence of soil pH on methane production. Williams & Crawford (1984) found that a pH increase from 3.2 to 5.8 increased the methane production of an incubated peat from a Minnesota peatland by 1.5 fold, for samples from 10 cm depth, and 2.2 fold for samples from 60 cm depth.
To complicate things further, some studies have reported negative relationships. Bergman et al (1999) reported a negative effect of pH on methane production in incubations of peat soil from a Swedish mire, but it was only statistically significant for one of two years
data. Valentine et al. (1994) also reported a decrease in methane production rates in peat with increasing pH. Bergman et al. (1998) reported both positive and negative relationships with pH for peat originating from different plant communities within the same mixed mire site in Sweden, and suggested that conflicting results may be due to competition for hydrogen between methanogens and homoacetogens, increasing pH favouring the later since it increases the degree of dissociation of acetate, or due to differences in cation exchange capacity ( CEC).
It is clear that the response of methane production to pH is very complex. Some of these conflicting results may have been confounded by methane oxidation. However, from the above results it can be stated that the rate of methane production is usually at an optimum ( m ///pH = 1) at around pH 7 (Garcia et al., 2000; Wang et al., 1993), and is close to the optimum between pH 5.5 -7 (Dunfield et al., 1993), decreasing to close to zero at around pH 3 (Williams & Crawford, 1984, 1985; Dunfield et al., 1993).
This response can be simulated using a sigmoid relationship, as
where pH is the measured pH of the soil layer, and c 3 and c 4 are constants ( c 3 = -1, and c 4 = -50). The relationship between m ///pH and pH is shown in Figure A2.13.
Figure A2.11 Response of anaerobic decomposition to soil pH
Soil factor for diffusion and oxidation of methane
Water table depth is one of the main factors controlling methane emissions as it determines the position of the boundary between the anaerobic and aerobic zones. When the water table is below the soil surface, oxidation of methane becomes a major controlling variable for methane efflux (Christensen et al., 2000). As a result, a lower water table decreases methane emission (Blodau et al., 2004) and draining peats may even convert the soil to a net methane sink (Blodau & Moore, 2003; Huttunen et al., 2003; Maljanen et al., 2002). Hargreaves & Fowler (1998) measured methane fluxes over a peat wetland in Caithness and related them to water table depths in different areas of the bog. They found a negative linear relationship between depths of around 8-17 cm. Daulat & Clymo (1998) also report a linear relationship between water table depth and methane emission, with depths of more than 15-20 cm below the soil surface stopping peat cores from being a net emitter of methane (Figure A2.12). MacDonald et al. (1998) found a reduction in methane emissions from Scottish peat cores with increasing water table depth, but in this case, the observed relationship is a decay curve and the cores do not become net sinks for methane even when the water table is 40 cm below the surface. Moore & Dalva (1993) found a negative logarithmic correlation with water table depth, down to 60 cm, in cores of peatland. Where the water table is above the surface, oxidation can also reduce methane emissions, particularly if light availability allows benthic photosynthetic activity (Le Mer & Roger, 2001).
Figure A2.12 Reported responses of methane emissions to water table depth
If it can be assumed that methane oxidation is close to zero when the water table is at the surface, the soil factor accounting for diffusion and oxidation of methane ( n) can be calculated from the depth (in cm) at which methane emissions cease and the soil becomes a net methane sink ( d sink) as follows:
This results in a value of v = 0.056 cm -1 for the soils of Daulat & Clymo (1998), and v = 0.04 cm -1 for the soils of Hargreaves and Fowler (1998). The methane emissions calculated in this way are shown in figure A2.13. Further work is required to determine the relationship between soil type and the depth at which the soil becomes a net methane sink.
Figure A2.13 Reduction in methane emissions with depth of water table, as calculated by ECOSSE
Methane oxidizing bacteria (methanotrophs) are more tolerant to pH variations than methanogens, with a reported optimum pH of 5.0 to 6.5 in temperate and subarctic peats (Dunfield et al., 1993), and have been discovered in peat soils below pH 4.7 using molecular ecological methods (McDonald et al., 1996). However, Hutsch et al. (1994) reported that, in a non-fertilised permanent grassland at the Rothamsted experimental station, a decrease in pH from 6.3 to 5.6 reduced methanotrophy by almost half. As a first approximation, it is assumed that methane oxidation is not influenced by soil pH.
A2.4 Description of dissolved organic matter (C and N): turnover and losses
Dissolved Organic Carbon ( DOC) is a potentially important component of carbon export from the soil system under highly organic conditions. Dissolved Organic Nitrogen ( DON) is closely linked to the production of DOC, and Jörgensen & Richter (1992) demonstrated that the two soil characteristics must be considered together in order to avoid unrealistic simulations of carbon or nitrogen concentration in the soil. Evans et al. (2005) and Evans et al. (in press) give a convincing argument that DOC production in organic soils is increasing, not due to climate change-induced oxidation, but instead as a result of soils recovering from acid deposition, which suppresses DOC solubility. They demonstrate a link between increased temperature, declining sulphur deposition and sea-salt loading. Strong links are already known to exist between soil water pH and DOC solubility, with an inverse relationship between mineral and organic acid export from soils (Krug & Frink, 1983). Palmer et al. (2001) used a combination of stable isotope and 14C-dating to identify the main sources and processes controlling DOC production in a temperate non-forested watershed underlain by mostly organic and podzolic soils. They found that the main source of DOC was the readily-decomposable carbon, such as leaf litter, rather than the longer term storage of SOM. They also showed that wetter soils export more 'recent' carbon than dry soils: DOC from dry soils contains more of the older carbon than does DOC from wet soils.
Dy DOC is a model of DOC production that utilises a layered soil description (Michalzik et al., 2003), and contains different organic fractions. Three processes are represented in Dy DOC: hydrology, metabolism and sorption. Hydrology is modelled in terms of macropores and micropores, between which dissolved organic carbon diffuses. Metabolism is modelled using relatively simple relationships to define the transformations between the three pools of organic carbon, with only temperature being a variable. Sorption occurs to and from soil solids from the different pools, and is controlled by DOC type, soil solution pH and cation content, and the nature of the soil solids (texture etc.). In ECOSSE, we follow a similar approach, simulating the hydrological, metabological and sorptive processes using the 5 soil organic matter pools: decomposable and resistant plant material, soil biomass, humus, and inert organic matter. A further organic matter pool is introduced, containing the DOC and DON that is in solution (see figure A2.14).
Figure A2.14 Pool structure for DOC production and uptake
This dissolved component is then susceptible to leaching by the same processes as already exist in the model to describe leaching losses. According to the model of Aguilar & Thibodeaux (2005), there are two fractions of DOC, one fraction that is readily available in the soil solution at all times, and another that is created from slower decomposition processes. This observation is consistent with the transfer into the DOC pool during decomposition of material from the rapidly turning over biomass and plant material pools, as well as from the more slowly turning over humus pool. In Dy DOC production by metabolism is calculated using a Q 10 relationship with a rate dependent only on temperature:
where k pool is the rate constant for DOC production specific to the pool; Q 10 is a constant, usually set to 2.0 ; T is the temperature (ºC); C pool is the amount of carbon in the pool (kg C ha -1) and t is the size of the timestep.
In ECOSSE we differentiate the equation, and include rate modifiers for moisture ( m //water), temperature ( m //temp), crop cover ( m //crop) and pH ( m //pH) as used in the existing calculation of SOM decomposition. The amount of DOC produced ( C poolÆ DOC) by a given pool is then calculated as:
where k pool is set to 0.0001 day -1 for decomposable plant material; 0.000005 day -1 for resistant plant material; 0.00005 day -1 for soil biomass; and 0.000002 day -1 for humus.
The moisture modifier ( m //water) is calculated from the available water at saturation 
, the available water at field capacity 
and the actual available water in the 5cm layer 
(all in mm) as follows:
The temperature rate modifier is calculated from the temperature of the soil layer in °C (T C) as follows:
The crop rate modifier is set as follows:
Finally, the pH rate modifier is calculated as discussed in section 2.3.7 as
where pH is the pH of the soil layer; and pH /min and pH /max are the pH values at which the minimum and maximum rate of decomposition occur respectively.
The carbon in the DOC pool ( C DOC) can further decompose to produce additional biomass ( C DOC?BIO) following a similar expression to the above:
where k DOC is set to 0.005 day -1.
The decomposition to biomass produces a fixed proportion of carbon dioxide ( C DOC?CO2), i.e.
where f eff is a fraction representing the efficiency of decomposition, and is set according to soil type to the same value as used in the decomposition routines. Having calculated the changes in carbon, the transformations in DON are calculated using the C:N ratio of the decomposing pool. Leaching of DOC takes place only for the portion that is not sorbed onto soil solids. The proportional availability ( DOCSOLV) of DOC is dependent on pH using the following relationship
DOCSOLV = C DOC ( pH - 3) / 4,
giving a maximum availability at pH = 7, and a minimum at pH = 3. If pH < 3 then DOCSOLV = 0, while if pH > 7 then DOCSOLV = C DOC.
A2.5 Improved layer structure in the soil profile
Figure A2.15 Distribution of ammonium, nitrate and soil organic matter in original model ( ECOSSE1) and model with improved layers ( ECOSSE2a)
In SUNDIAL- MAGEC, the soil is divided into different layers for the different components of the soil: ammonium is simulated in 50cm layers down the soil profile; SOM is simulated in 25cm layers to a maximum depth of 50cm; and nitrate is simulated in 5cm layers for the top 50cm, followed by 50cm layers for the remainder of the profile. This structure does not allow C and N turnover in deep organic soils to be simulated. In ECOSSE, the layers are unified into 5cm layers throughout (Figure A2.15), so allowing C and N turnover to be simulated to depth. The soil parameters are also described in 5cm layers, so allowing more detailed descriptions of soil profile characteristics to be included. This is important in highly organic soils where the organic layer can be very deep.
A2.6. Initialisation of the size and characteristics of the soil organic matter pools
In SUNDIAL- MAGEC, the status of the SOM at the start of the simulation is entered as a soil parameter. These parameters describe not only the total amount of organic matter in the soil, but also its activity, which is an important driver for the amount of C and N released from the organic matter during the simulation. For mineral soils, these parameters have been derived from measurements of soil C in long-term experiments, for which average climate and long-term land use can be estimated. The standard parameters derived for mineral soils are not transferable to organic soils. Furthermore, organic soils contain a higher percentage of SOM, so the influence of measured C content on C and N release is more pronounced. It is not likely that fixed soil parameters describing amount and activity of SOM will provide an accurate simulation of C and N release in organic soils: these factors should instead be entered as input variables that change for each simulation.
The developments contained in ECOSSE allow the amount and activity of SOM to be estimated using two additional input variables: the dominant land use over the last 10,000 years, and the measured soil C content at the start of the simulation. If it can be assumed that the soil is close to its equilibrium SOM content, the activity of the SOM can be estimated using an equilibrium run of ROTH-C (Coleman & Jenkinson, 1996, Figure A2.16). The equilibrium run has been built into the initialisation routines of ECOSSE.
Figure A2.16 Initialisation of amount and activity of soil organic matter
The dominant land use determines the pattern of plant inputs used in the ROTH-C equilibrium run. The distributions used, obtained from Smith et al. (2005), are shown in Figure 2.20. These distributions are used as the default monthly plant input of C in a first equilibrium run. The equilibrium run calculates the amount of C in each SOM pool. The total C at equilibrium can be obtained by summing the C content of these pools. The annual input of plant C needed to achieve the measured soil C at equilibrium can then be estimated from the ratio of the measured and simulated soil C and the default plant input used in the run (Figure A2.17). The estimated annual plant C input is then redistributed as specified by the dominant land use (Figure A2.17), and used in a further equilibrium run. This run, calculates the amount of C in each SOM pool, given the new estimate of input plant C. The annual input of plant C needed to achieve the measured soil C at equilibrium is then re-estimated from the measured and new simulated soil C. This procedure is continued until the measured and simulated values of soil C differ by less than 0.00001 t C ha -1. The final estimates of monthly inputs of plant C and the amount of C in each SOM pool are then used in the full simulation.
Figure A2.17 Distribution of plant inputs for 3 major land use
A2.7 Incorporation of the effect of pH on soil processes
pH effect on rate constant: A significant effect of soil pH on the rate of decomposition has been observed in many studies (e.g. Andersson and Nilsson, 2001; Hall et al. 1997; Situala et al. 1995). High soil acidity is generally considered to limit the activity of decomposers but studies that have manipulated pH in the field or laboratory have found conflicting results. This can be explained by the interaction between soil pH, organic matter decomposition and nutrient cycling (Binkley and Richter, 1987).
Motavalli et al. (1995) found a positive correlation between biological activity and soil pH in soils amended with 14C labelled plant residues. Similarly, Sitaula et al. (1995) used acid irrigation to examine the effect of low pH and reported that pH 3 produced CO 2 fluxes 20 % lower than pH 4 and 5.5, between which there was no significant difference. Persson and Wiren (1989) reported increasing the acidity of forest soil from pH 3.8 to 3.4 reduced CO 2 evolution by 83 % and from pH 4.8 to 4 by 78 %.
However, these effects differ with redox conditions. Increases in pH have been reported to increase CO 2 production 1.4-fold under anaerobic conditions but decrease it by 53 % under aerobic conditions (Bridgham and Richardson 1992). Bergman et al. (1999) compared CO 2 production rates at pH 4.3 and 6.2, and found that under anaerobic conditions rates were 21 (at 7 oC) and 29 (at 17 oC) times greater at the more neutral pH, while under aerobic conditions rates were 3 times greater at 7 oC but pH had no significant effect at 17 oC.
In a simple regression model, Reth et al. (2005) included the effect of soil pH in terms of the deviation of the soil pH ( pH) from the optimum pH for decomposition ( pH opt), and the sensitivity of the decomposition processes in this soil to pH ( pH sens):
The plot in Figure A2.18 shows how Reth's model can be used to simulate the aerobic observations described above.
Figure A2.18 Fitted equation by Reth et al. (2005) to the observations of changes in aerobic decomposition with pH made by Situala et al. (1995) and Persson & Wiren (1989).
The plots show that Reth's approach can be used to accurately describe the response of aerobic decomposition to pH, if the optimum pH for decomposition and the sensitivity to pH are known, and the response is assumed to flatten off above the optimum pH and below a minimum pH. The difficulty in implementing this in a functional model is in knowing what these values should be.
The influence of soil pH on decomposition is not implemented in the SUNDIAL model as it was originally designed to work in well-managed arable soils, and so it could be assumed that the pH was close to neutral (Bradbury et al., 1993). This assumption breaks down in natural and managed highly organic soils, where the pH is more variable. The above discussion suggests that the implementation of decomposition in ECOSSE should include a different description of the effects of pH on aerobic and anaerobic decomposition. In an approach that follows that of Reth et al., 2005, but with a simplified formula that uses more explicit terms, aerobic decomposition is described as proceeding at an optimum rate (rate modifier W pH = 1) until the pH falls below a critical threshold ( pH max), after which the rate of decomposition falls to a minimum rate (rate modifier W pH = W pH,,min) at pH min (see Figure A2.19a):
Similarly, anaerobic decomposition should proceed at an optimum rate (rate modifier W pH/ = 1) until the pH falls below a critical threshold ( pH max/), after which the rate of decomposition falls to a minimum rate (rate modifier W pH/ = W pH,,min/) at pH min/ (see Figure A2.19b):
That these much simpler equations can equally be used to describe the observations of aerobic and anaerobic decomposition is demonstrated in figure A2.20. In order to use these equations in a range of soil types, experimental data is needed to obtain generic parameters for the soils according to texture and organic matter content (as required in large scale simulations). Alternatively, observations of changes in CO 2 production with pH could be used to set the parameters at a specific site.
Figure A2.19 Decomposition rate modifiers for pH; (a) W pH = aerobic rate modifier; (b) W pH / = anaerobic rate modifier.
Figure A2.20 ECOSSE equation for effect of soil pH on rate constant, fitted to (a) changes in aerobic decomposition with pH (Situala et al., 1995; Persson & Wiren, 1989); (b) changes in anaerobic decomposition with pH made (Bergman et al., 1999)
pH effect on stable C:N ratio: Change in pH has also been observed to result in a change in the C:N ratio of the organic matter in the soil. The stable C:N ratio is an important driver of mineralization or immobilisation of N during the decomposition of organic matter. In SUNDIAL- MAGEC, the stable C:N ratio of SOM is assumed to be 8 (Bradbury et al., 1987). If organic matter has a C:N ratio greater than 8, then the organic matter is deficient in C, and N will be mineralised to ammonium during decomposition. If organic matter has a C:N ratio less than 8, then the organic matter is deficient in N, and N will be immobilised during decomposition, first from the ammonium, and then from the nitrate pool.
The choice of a stable C:N ratio of 8 is a result of the weighted average of the C:N ratios of the decomposing fungi and bacteria in the soil. However, as the soil pH falls, the ratio of fungi to bacteria increases because fungi are more tolerant of acidic conditions. Because the C:N ratio of fungi is higher than the C:N ratio of bacteria, this change in the population is accompanied by an increase in the stable C:N ratio of the SOM.
This is due to the decomposer community in the soil becoming dominated by fungi with a higher C:N ratio than bacteria. This suggests that the stable C:N ratio of the organic matter pools in ECOSSE should change with changing pH. Again, in the absence of more detailed experimental data, a simple linear approach is used (see Figure A2.21).
The proportion of bacteria in the soil, p bac, is calculated as
where b min = 0.2 is the minimum proportion of bacteria found in the soil, b max = 0.5 is the maximum proportion of bacteria found in the soil, pH is the pH of the soil, pH min is the pH at which minimum soil bacteria occurs, and pH max is the pH at which maximum soil bacteria occurs.
Figure A2.21 Calculation of change in stable C:N ratio with pH; (a) proportion of fungi ( p F) and bacteria ( p B); (b) stable C:N ratio.
Similarly, the proportion of fungi in the soil, p fun, is calculated as
The stable C:N ratio of the SOM, ( C : N) stable, can then be calculated from the proportion of bacteria and fungi in the soil, and the typical C:N ratios of bacteria and fungi, ( C : N) bac = 5.5 and ( C : N) fun = 11.5 respectively, i.e.
The change in stable C:N ratio of SOM between pH 5.5 and 4 is shown in Figure A2.22.
Figure A2.22 Change in stable C:N ratio with pH
A2.8 Incorporation of the effect of saturated conditions on soil processes
There are a number of different ways to express soil water content. Volumetric water content (% or mm/mm) gives a good indication on how much water is there, but to assess the effect on soil processes also soil specific parameters must be taken into account. There are 3 descriptors of soil water behaviour that are needed: permanent wilting point, field capacity and saturation. Permanent wilting point is the limit below which plants can extract no more water from the soil. Field capacity is the maximum water that the soil can hold against gravity, and therefore the maximum soil water content that can occur in a freely drained soil. As agricultural soils are usually freely drained, agricultural models usually ignore water contents above field capacity. The real maximum water content of a soil is called saturation. That is the maximum amount of water that can fit in the soil before it overflows, when all pores are filled with water.
Field capacity is generally assumed to be the optimum water content for biological processes because both water and oxygen are easily available. Below field capacity water availability may be limiting; above field capacity oxygen availability may be limiting. This can lead to reduction of the rate of some processes, but also to whole new processes taking place (such as methane and nitrous oxide formation) as electron acceptors other than oxygen are used.
In models where the purpose is to express possible water limitation, water content is usually expressed as available water, as is the case in ECOSSE.
When water content < permanent wilting point, available water = 0.
When water content > permanent wilting point, available water = water content - permanent wilting point.
In models where the purpose is to predict methane emissions and denitrification, water contents above field capacity must also be considered. Whilst methane emissions occur in strongly anoxic soil, denitrification occurs at water contents below field capacity and increases with water content. The water content is more usually expressed as the percentage of soil pores filled with water.
To use equations from other models for the effect of soil water content on nitrous oxide emissions it is necessary to calculate water filled pore space from water content and vice versa. Water filled pore space can be expressed as:
where w is water filled pore space, vw is volume of water and wp is total pore volume. The total pore volume can be calculated as: pore volume = total volume - volume of solid material. For soils, that simplifies to:
where S t = total soil porosity, b = bulk density and p = particle density (Danielson & Sutherland, 1986). As most mineral soils are from similar material ("rock"), particle density does not vary much between mineral soils, and is usually close to 2.65 g cm -3. The measured values for peat are somewhat lower: Pihlatie et al. (2004) measured a soil porosity for peat of 1.9 g cm -3. If saturation water content is known, water filled pore space will be the percentage of that water content, as saturation water content equals the total pore volume filled. To calculate saturation water content, bulk density and particle density must be known.
Another way to express oxygen limitation is redox potential (Li, 2000). The advantage of this measure is that the effect of excess water can be integrated over time: When the soil becomes saturated, first stored oxygen will be used up. After that, other electron acceptors are used in a sequence depending on their redox potential. One of the effects of this is that methane emissions usually only starts after the soil has been saturated for weeks. This is difficult to simulate if only the water content in the individual time step is input. A simplification is to introduce a time lag in the onset of methane emissions.
Most of the models that do incorporate an effect of water saturation on decomposition simply decrease decomposition rate of all pools equally (Parton et al., 1994). Whilst this can be tuned to fit most data, there is experimental evidence that that a stronger effect on resistant pool may be more mechanistically correct. Lignin, the most resistant material does not decompose at all under anaerobic conditions, and cellulose decomposes only slowly (Kögel-Knabner, 2002). Litter quality is also the main determinant of decomposition rate under anaerobic conditions (Miyajima et al., 1997). There are also reports that sugar and easily decomposable material decomposes at about the same rate under anaerobic as under aerobic condition (Bergman et al., 1999), but with a larger portion of carbon is released as CH 4 rather than CO 2. Therefore, the reduction in decomposition rate under anaerobic conditions should be implemented most strongly in the RPM pool. The implication of this will be that RPM will accumulate under anaerobic conditions, as is expected during peat accumulation. That means that if conditions become aerobic, this organic matter will decompose quickly. Again, this agrees with observations. Figure A2.23 shows the structure of the model.
Figure A2.23 Diagram of how anaerobic decomposition can be integrated into ECOSSE soil organic matter model. The upper pathway shows decomposition of easily decomposable parts of the litter (Shirato & Yokozawa, 2006). This is envisaged to occur both at anaerobic and aerobic conditions, but under anaerobic conditions some of the carbon is released as methane. When the microbes die, parts of the dead cells are decomposed, but parts are left in the soil. This forms the majority of soil organic matter under aerobic conditions. The lower pathway shows decomposition of more resistant material (Shirato& Yokozawa, 2006). The microbes dealing with these processes are severely limited under anaerobic conditions, therefore recalcitrant material is left un-decomposed or partly decomposed. This will form the bulk of soil organic matter under anaerobic conditions.
To obtain parameters for the model, data were used that had been collected from incubation experiments where flooded and non-flooded decomposition was compared (Figure A2.24). An incubation experiment with rice straw indicated that the mass loss over 160 days under flooded conditions was about 80% of that at 50% WHC (Devevre & Horwarth, 2000). This experiment was used to fit the water modifier to the decomposition rate. Measurements of respiration in peat soils, which can be expected to contain more DPM, were more limited by anaerobic condition, 34-62% of aerobic (Bridgeham & Richardson, 1992), so the modifier for RPM was set to a lower level (0.2), and the modifier for DPM was then fitted. It was assumed that the modifier for biomass and DPM were the same, as well as the modifier for RPM and humus. The fitting was done not only to optimize fit, but also to make any discrepancies in temperature response similar at flooded and non-flooded conditions. It is difficult to say at what water-filled pore space decomposition rate should start to decline except that it should be at water contents above field capacity. We therefore assume a linear decline from field capacity to saturation.
Figure A2.24 Measured and simulated graphs for mass loss during incubation of rice straw under flooded and non-flooded conditions (Devevre & Horwarth, 2000). The water modifier is set at 1 for the non-flooded treatments and fitted for the flooded treatment.
« Previous | Contents | Next »