Models that contain several measurable compartments should be simpler to parameterise and verify, particularly if the dynamics of the fractions they simulate can be measured on a realistic time scale. Furthermore, model parameterisation will be more robust and straightforward, agreement having to be reached between the dynamic, simultaneous measurement of several variables. The predictions of the parameterised model could also be verified by monitoring the concentration of the various SOM fractions, and initialisation could be confirmed by measurement. A reliance on empirically derived rules (which require detailed site-specific information) should be reduced.
A simple experimental situation used to parameterise such a model would involve the incorporation of crop residues into soil, followed by incubation (possibly under controlled conditions), and fractionation to determine the size of model compartments over time. Once developed for use in complex systems, the model could be verified in the field. Long-term data would still be required to determine the turnover of the less reactive SOM fractions, but the dynamics of several key compartments would provide good indication of likely applicability, and rigorous parameterisation would be required only over the time period relevant to the application. Providing the model could simulate the fractions dynamic on the time scale relevant to the application, the background activity of stable (less reactive) fractions could be considered constant.
A model parameterised using relatively short-term data provide could provide adequate prediction of soil N supply.
1.4.1. Operationally-defined fractions for modelling
Since it is optimistic to expect an experimentally obtained SOM fraction to be perfectly defined with respect to a pre-determined reactivity (k), it is more realistic to consider them defined by a SOM fractionation procedure. These operationally- (rather than conceptually- defined SOM fractions are not only measurable, but may be isolated by physical rather than chemical fractionation, and thus reflect the influence of their location in the soil rather than their chemical reactivity per se.
Conceptually, current models (with exceptions e.g. Verberne et al. 1990) assume compartment reactivity is governed solely by chemical composition. Their reliance on texture modifiers (usually based on soil clay content) may reflect the lack of an explicit consideration of soil structure (e.g. physical protection of SOM in aggregates) as much as the functional characteristics of clay itself (McGill 1996). These modifiers are generally applied to the compartment rate constants (Molina and Smith 1997), but may also alter flow-partitioning parameters according to empirically derived equations e.g. CENTURY (Parton 1996). Evidence increasingly supports, however, the view that physical effects are of equal importance to SOM turnover as chemical composition, and should therefore be explicitly represented in models e.g. Balesdent (1996).
A conceptual model based on physical location is highly compatible with physical fractionation i.e. separation of SOM by particle-density and / or particle-size
(Christensen 1996b). Since physical fractionation is relatively non-destructive, they offer possibilities for the chemical characterisation of modelled fractions, and hence verification at a process-level.
A potential drawback in modelling the measurable, however, is that measured fractions may be seen to vary in composition, and potentially reactivity, over time. To accommodate these observations new, more radical mathematical frameworks may be required, possibly allowing time-dependent variation in k.
1.4.2. Physical fractionation
Physical fractionation is intended to emphasise the close relationship between the physical location of SOM and its decomposition. However, there is no presumption as to whether any observed differences in chemical composition are conferred by location, or vice versa. On the basis of the main separation properties available – particle-density and particle-size – conceptual categorisations can be made. On the basis of density SOM can be divided into light and heavy fractions, comprising organic particles and aggregated mineral particles respectively. Organic particles located within aggregates can be isolated as light material, if aggregates are first broken down into primary particles (Gregorich and Janzen 1996). Protected within aggregates, it seems such particles differ in decomposition rate (Gregorich et al. 1996), and this may be reflected in their chemical composition (Golchin et al. 1994b). The protected (or occluded / intra-aggregate) fraction may slowly form around residues of primary substrates, or around centres of microbial activity during substrate utilisation (Spycher
organic macromolecules to form organomineral complexes, products of the interaction between slowly decomposing organic material and microbial enzyme activity (Christensen 1996a).
Clay-size organomineral complexes may have a particular significance in the turnover of SOM (Christensen 1992), and can be isolated (after soil dispersion) using simple methods based on sedimentation. The settling velocity (s) of a spherical particle in suspension is determined by Stokes Law:
s =
[
G.D2(dρ – dl )]
/ 18n– where G is the acceleration induced by gravity, D the diameter of the particle, dρ its density, and d1 and n the density and viscosity of the suspension (the latter being temperature dependent). The calculated settling rate is highly sensitive to particle diameter, although the assumption of spherical particles may be unrealistic for laminar clays.
Physical fractionation offers a new conceptual model for SOM decomposition, in which free light particles comprise the most reactive fraction, protected light particles a less reactive fraction, and organomineral material the least reactive (possibly varying with particle-size). It may be envisaged that plant inputs pass – with successive transformations through microbial biomass – from the free light to protected light fraction and / or organomineral material. If the distribution of microbial biomass were determined, it would likely be located within all three of these fractions.
The chemical composition of a SOM fraction is not likely to be greatly altered during separation, and non-invasive characterisation should reflect the composition of the material available to decomposer organisms in the soil. Physical protection may be measured by comparing the reactivity of a fraction in situ with that measured after isolation.