When is a measured soil organic matter fraction equivalent to a model pool?

Dynamic simulation models are important tools for rationalizing complex changes in soil organic matter. Most such models for organic matter can be described as having either a pool structure or a continuous one. A pool structure can offer advantages in ease of use and transferability. Some pools are easily measured, whereas others cannot be measured directly. New methods of fractionation are being developed in an attempt to base models on measurable fractions. A requirement for such models is a demonstration that the measured fraction and model pool are equivalent. A measured fraction is equivalent to a model pool only if, within acceptable limits, it is unique as well as non-composite. If the measured fraction is not unique, describing it as a separate pool adds no extra information, while the added complexity can increase propagation of errors. If it is composite then the characteristics of the fraction will change with changing environment as a result of changing proportions of subpools. This will produce a model that cannot be applied without deriving parameters afresh: such a model is of greatly reduced value. Here we develop methods to examine if a fraction is both unique and non-composite. The tests for unique and composite pools were applied to the SUNDIAL (simulation of nitrogen dynamics in arable land) model of organic matter and nitrogen turnover in soil. Results suggest that the debris, biomass and humus pools are unique, but biomass and humus are composed of two or more subpools. This worked example illustrates how, given suitable data, any pool-based model can be tested by these methods.